use Venice context tier thresholds

feat(chutes): add API-driven model generator script

bun.lock

@@ -32,7 +32,6 @@
     "packages/web": {
       "name": "@models.dev/web",
       "dependencies": {
-        "@tanstack/virtual-core": "^3.14.0",
         "hono": "^4.8.0",
         "models.dev": "workspace:*",
       },
@@ -58,8 +57,6 @@
 
     "@models.dev/web": ["@models.dev/web@workspace:packages/web"],
 
-    "@tanstack/virtual-core": ["@tanstack/virtual-core@3.14.0", "", {}, "sha512-JLANqGy/D6k4Ujmh8Tr25lGimuOXNiaVyXaCAZS0W+1390sADdGnyUdSWNIfd49gebtIxGMij4IktRVzrdr12Q=="],
-
     "@tsconfig/bun": ["@tsconfig/bun@1.0.8", "", {}, "sha512-JlJaRaS4hBTypxtFe8WhnwV8blf0R+3yehLk8XuyxUYNx6VXsKCjACSCvOYEFUiqlhlBWxtYCn/zRlOb8BzBQg=="],
 
     "@types/bun": ["@types/bun@1.3.0", "", { "dependencies": { "bun-types": "1.3.0" } }, "sha512-+lAGCYjXjip2qY375xX/scJeVRmZ5cY0wyHYyCYxNcdEXrQ4AOe3gACgd4iQ8ksOslJtW4VNxBJ8llUwc3a6AA=="],

package.json

@@ -17,6 +17,7 @@
   "scripts": {
     "validate": "bun ./packages/core/script/validate.ts",
     "compare:migrations": "bun ./packages/core/script/compare-model-migrations.ts",
+    "chutes:generate": "bun ./packages/core/script/generate-chutes.ts",
     "helicone:generate": "bun ./packages/core/script/generate-helicone.ts",
     "venice:generate": "bun ./packages/core/script/generate-venice.ts",
     "vercel:generate": "bun ./packages/core/script/generate-vercel.ts",

packages/core/script/generate-chutes.ts

@@ -0,0 +1,589 @@
+#!/usr/bin/env bun
+
+/**
+ * Generates Chutes model TOML files from the Chutes LLM API.
+ *
+ * Flags:
+ * --dry-run: Preview changes without writing files
+ * --new-only: Only create new models, skip updating existing ones
+ * --keep-orphans: Don't delete TOML files for models no longer in the API
+ */
+
+import { z } from "zod";
+import path from "node:path";
+import { mkdir } from "node:fs/promises";
+import { ModelFamilyValues } from "../src/family.js";
+
+const API_ENDPOINT = "https://llm.chutes.ai/v1/models";
+
+enum SkipZeroFields {
+  LimitContext = "limit.context",
+  LimitOutput = "limit.output",
+}
+
+const Pricing = z.object({
+  prompt: z.number().optional(),
+  completion: z.number().optional(),
+  input_cache_read: z.number().optional(),
+}).passthrough();
+
+const ChutesModel = z.object({
+  id: z.string(),
+  created: z.number(),
+  pricing: Pricing.optional(),
+  context_length: z.number().optional(),
+  max_output_length: z.number().optional(),
+  max_model_len: z.number().optional(),
+  input_modalities: z.array(z.string()).optional(),
+  output_modalities: z.array(z.string()).optional(),
+  supported_features: z.array(z.string()).optional(),
+  supported_sampling_parameters: z.array(z.string()).optional(),
+  quantization: z.string().optional(),
+}).passthrough();
+
+const ChutesResponse = z.object({
+  data: z.array(ChutesModel),
+}).passthrough();
+
+interface ExistingModel {
+  name?: string;
+  family?: string;
+  attachment?: boolean;
+  reasoning?: boolean;
+  tool_call?: boolean;
+  structured_output?: boolean;
+  temperature?: boolean;
+  knowledge?: string;
+  release_date?: string;
+  last_updated?: string;
+  open_weights?: boolean;
+  interleaved?: boolean | { field: string };
+  status?: string;
+  cost?: {
+    input?: number;
+    output?: number;
+    cache_read?: number;
+  };
+  limit?: {
+    context?: number;
+    output?: number;
+  };
+  modalities?: {
+    input?: string[];
+    output?: string[];
+  };
+}
+
+interface MergedModel {
+  name: string;
+  family?: string;
+  attachment: boolean;
+  reasoning: boolean;
+  tool_call: boolean;
+  structured_output?: boolean;
+  temperature: boolean;
+  knowledge?: string;
+  release_date: string;
+  last_updated: string;
+  open_weights: boolean;
+  interleaved?: boolean | { field: string };
+  status?: string;
+  cost?: {
+    input: number;
+    output: number;
+    cache_read?: number;
+  };
+  limit: {
+    context: number;
+    output: number;
+  };
+  modalities: {
+    input: string[];
+    output: string[];
+  };
+}
+
+interface Changes {
+  field: string;
+  oldValue: string;
+  newValue: string;
+}
+
+// ── Utility functions ────────────────────────────────────────────────
+
+function timestampToDate(timestamp: number): string {
+  const date = new Date(timestamp * 1000);
+  return date.toISOString().slice(0, 10);
+}
+
+function getTodayDate(): string {
+  return new Date().toISOString().slice(0, 10);
+}
+
+function formatNumber(n: number): string {
+  if (n >= 1000) {
+    return n.toString().replace(/\B(?=(\d{3})+(?!\d))/g, "_");
+  }
+  return n.toString();
+}
+
+/**
+ * Humanize a model ID into a readable name.
+ * Strips the org prefix and replaces hyphens with spaces.
+ * e.g. "Qwen/Qwen3-32B-TEE" → "Qwen3 32B TEE"
+ */
+function humanizeModelName(modelId: string): string {
+  const parts = modelId.split("/");
+  const modelPart = parts[parts.length - 1];
+  return modelPart.replace(/-/g, " ");
+}
+
+// ── Family inference (same approach as generate-vercel.ts) ───────────
+
+function isSubstring(target: string, family: string): boolean {
+  return target.toLowerCase().includes(family.toLowerCase());
+}
+
+function matchesFamily(target: string, family: string): boolean {
+  const targetLower = target.toLowerCase();
+  const familyLower = family.toLowerCase();
+  let familyIdx = 0;
+
+  for (let i = 0; i < targetLower.length && familyIdx < familyLower.length; i++) {
+    if (targetLower[i] === familyLower[familyIdx]) {
+      familyIdx++;
+    }
+  }
+
+  return familyIdx === familyLower.length;
+}
+
+function inferFamily(modelId: string, modelName: string): string | undefined {
+  const sortedFamilies = [...ModelFamilyValues].sort((a, b) => b.length - a.length);
+
+  // First pass: try exact substring matches
+  for (const family of sortedFamilies) {
+    if (isSubstring(modelId, family)) {
+      return family;
+    }
+  }
+
+  for (const family of sortedFamilies) {
+    if (isSubstring(modelName, family)) {
+      return family;
+    }
+  }
+
+  // Second pass: fall back to subsequence matching
+  for (const family of sortedFamilies) {
+    if (matchesFamily(modelId, family)) {
+      return family;
+    }
+  }
+
+  for (const family of sortedFamilies) {
+    if (matchesFamily(modelName, family)) {
+      return family;
+    }
+  }
+
+  return undefined;
+}
+
+// ── Load existing TOML ───────────────────────────────────────────────
+
+async function loadExistingModel(filePath: string): Promise<ExistingModel | null> {
+  try {
+    const file = Bun.file(filePath);
+    if (!(await file.exists())) {
+      return null;
+    }
+    const toml = await import(filePath, { with: { type: "toml" } }).then(
+      (mod) => mod.default,
+    );
+    return toml as ExistingModel;
+  } catch (e) {
+    console.warn(`Warning: Failed to parse existing file ${filePath}:`, e);
+    return null;
+  }
+}
+
+// ── Merge API data with existing TOML ────────────────────────────────
+
+function mergeModel(
+  apiModel: z.infer<typeof ChutesModel>,
+  existing: ExistingModel | null,
+): MergedModel {
+  const features = new Set(apiModel.supported_features ?? []);
+  const samplingParams = new Set(apiModel.supported_sampling_parameters ?? []);
+  const inputMods = apiModel.input_modalities ?? ["text"];
+  const outputMods = apiModel.output_modalities ?? ["text"];
+
+  // Capabilities from API features
+  const hasAttachment = inputMods.some((m) =>
+    m === "image" || m === "video" || m === "pdf",
+  );
+  const hasReasoning = features.has("reasoning");
+  const hasToolCall = features.has("tools");
+  const hasStructuredOutput = features.has("structured_outputs");
+  const hasTemperature = samplingParams.size > 0
+    ? samplingParams.has("temperature")
+    : true; // default true if no sampling params info
+
+  // Preserve existing values when available (manually specified)
+  const modelName = existing?.name ?? humanizeModelName(apiModel.id);
+  const family = existing?.family ?? inferFamily(apiModel.id, modelName);
+  const knowledge = existing?.knowledge;
+  const interleaved = existing?.interleaved;
+  const status = existing?.status;
+
+  // Release date: existing > API created timestamp > today
+  const releaseDate = existing?.release_date
+    ?? timestampToDate(apiModel.created)
+    ?? getTodayDate();
+
+  // Context limit: prefer context_length, fallback to max_model_len
+  const apiContext = apiModel.context_length ?? apiModel.max_model_len ?? 0;
+  const contextLimit = apiContext > 0
+    ? apiContext
+    : (existing?.limit?.context ?? 0);
+
+  // Output limit: prefer max_output_length, fallback to existing
+  const apiOutput = apiModel.max_output_length ?? 0;
+  const outputLimit = apiOutput > 0
+    ? apiOutput
+    : (existing?.limit?.output ?? 0);
+
+  const merged: MergedModel = {
+    name: modelName,
+    family,
+    attachment: hasAttachment,
+    reasoning: hasReasoning,
+    tool_call: hasToolCall,
+    temperature: hasTemperature,
+    release_date: releaseDate,
+    last_updated: getTodayDate(),
+    open_weights: true, // Chutes hosts open-weight models
+    ...(hasStructuredOutput && { structured_output: hasStructuredOutput }),
+    ...(knowledge && { knowledge }),
+    ...(interleaved !== undefined && { interleaved }),
+    ...(status && { status }),
+    limit: {
+      context: contextLimit,
+      output: outputLimit,
+    },
+    modalities: {
+      input: inputMods,
+      output: outputMods,
+    },
+  };
+
+  // Cost: API values are already in USD per 1M tokens — use directly
+  if (apiModel.pricing) {
+    const inputPrice = apiModel.pricing.prompt;
+    const outputPrice = apiModel.pricing.completion;
+    const cacheReadPrice = apiModel.pricing.input_cache_read;
+
+    if (inputPrice !== undefined && outputPrice !== undefined) {
+      merged.cost = {
+        input: inputPrice,
+        output: outputPrice,
+        ...(cacheReadPrice !== undefined && { cache_read: cacheReadPrice }),
+      };
+    }
+  }
+
+  return merged;
+}
+
+// ── TOML formatting ──────────────────────────────────────────────────
+
+function formatToml(model: MergedModel): string {
+  const lines: string[] = [];
+
+  lines.push(`# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.`);
+  lines.push(`# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status`);
+  lines.push(`name = "${model.name.replace(/"/g, '\\"')}"`);
+  if (model.family) {
+    lines.push(`family = "${model.family}"`);
+  }
+  lines.push(`release_date = "${model.release_date}"`);
+  lines.push(`last_updated = "${model.last_updated}"`);
+  lines.push(`attachment = ${model.attachment}`);
+  lines.push(`reasoning = ${model.reasoning}`);
+  lines.push(`temperature = ${model.temperature}`);
+  lines.push(`tool_call = ${model.tool_call}`);
+  if (model.structured_output !== undefined) {
+    lines.push(`structured_output = ${model.structured_output}`);
+  }
+  lines.push(`open_weights = ${model.open_weights}`);
+  if (model.knowledge) {
+    lines.push(`knowledge = "${model.knowledge}"`);
+  }
+  if (model.status) {
+    lines.push(`status = "${model.status}"`);
+  }
+
+  if (model.cost) {
+    lines.push("");
+    lines.push(`[cost]`);
+    lines.push(`input = ${model.cost.input}`);
+    lines.push(`output = ${model.cost.output}`);
+    if (model.cost.cache_read !== undefined) {
+      lines.push(`cache_read = ${model.cost.cache_read}`);
+    }
+  }
+
+  lines.push("");
+  lines.push(`[limit]`);
+  lines.push(`context = ${formatNumber(model.limit.context)}`);
+  lines.push(`output = ${formatNumber(model.limit.output)}`);
+
+  lines.push("");
+  lines.push(`[modalities]`);
+  lines.push(`input = [${model.modalities.input.map((m) => `"${m}"`).join(", ")}]`);
+  lines.push(`output = [${model.modalities.output.map((m) => `"${m}"`).join(", ")}]`);
+
+  if (model.interleaved !== undefined) {
+    lines.push("");
+    if (model.interleaved === true) {
+      lines.push(`interleaved = true`);
+    } else if (typeof model.interleaved === "object") {
+      lines.push(`[interleaved]`);
+      lines.push(`field = "${model.interleaved.field}"`);
+    }
+  }
+
+  return lines.join("\n") + "\n";
+}
+
+// ── Change detection ─────────────────────────────────────────────────
+
+function detectChanges(
+  existing: ExistingModel | null,
+  merged: MergedModel,
+): Changes[] {
+  if (!existing) return [];
+
+  const changes: Changes[] = [];
+  const EPSILON = 0.001;
+
+  const shouldSkipZero = (field: string, oldVal: unknown, newVal: unknown): boolean => {
+    if (!Object.values(SkipZeroFields).includes(field as SkipZeroFields)) {
+      return false;
+    }
+    return (typeof oldVal === "number" && oldVal === 0) || (typeof newVal === "number" && newVal === 0);
+  };
+
+  const formatValue = (val: unknown): string => {
+    if (typeof val === "number") return formatNumber(val);
+    if (Array.isArray(val)) return `[${val.join(", ")}]`;
+    if (val === undefined) return "(none)";
+    return String(val);
+  };
+
+  const isMaterialPriceDiff = (oldPrice: unknown, newPrice: unknown): boolean => {
+    if (oldPrice === 0 && newPrice === undefined) return false;
+    if (oldPrice !== undefined && newPrice !== undefined) {
+      return Math.abs((oldPrice as number) - (newPrice as number)) > EPSILON;
+    }
+    return oldPrice !== newPrice;
+  };
+
+  const compare = (field: string, oldVal: unknown, newVal: unknown) => {
+    if (shouldSkipZero(field, oldVal, newVal)) return;
+
+    const isDiff = field.startsWith("cost.")
+      ? isMaterialPriceDiff(oldVal, newVal)
+      : JSON.stringify(oldVal) !== JSON.stringify(newVal);
+
+    if (isDiff) {
+      changes.push({
+        field,
+        oldValue: formatValue(oldVal),
+        newValue: formatValue(newVal),
+      });
+    }
+  };
+
+  compare("name", existing.name, merged.name);
+  compare("family", existing.family, merged.family);
+  compare("attachment", existing.attachment, merged.attachment);
+  compare("reasoning", existing.reasoning, merged.reasoning);
+  compare("tool_call", existing.tool_call, merged.tool_call);
+  compare("structured_output", existing.structured_output, merged.structured_output);
+  compare("open_weights", existing.open_weights, merged.open_weights);
+  compare("release_date", existing.release_date, merged.release_date);
+  compare("cost.input", existing.cost?.input, merged.cost?.input);
+  compare("cost.output", existing.cost?.output, merged.cost?.output);
+  compare("cost.cache_read", existing.cost?.cache_read, merged.cost?.cache_read);
+  compare("limit.context", existing.limit?.context, merged.limit.context);
+  compare("limit.output", existing.limit?.output, merged.limit.output);
+  compare("modalities.input", existing.modalities?.input, merged.modalities.input);
+  compare("modalities.output", existing.modalities?.output, merged.modalities.output);
+
+  return changes;
+}
+
+// ── Main ─────────────────────────────────────────────────────────────
+
+async function main() {
+  const args = process.argv.slice(2);
+  const dryRun = args.includes("--dry-run");
+  const newOnly = args.includes("--new-only");
+  const keepOrphans = args.includes("--keep-orphans");
+
+  const modelsDir = path.join(
+    import.meta.dirname,
+    "..",
+    "..",
+    "..",
+    "providers",
+    "chutes",
+    "models",
+  );
+
+  console.log(`${dryRun ? "[DRY RUN] " : ""}${newOnly ? "[NEW ONLY] " : ""}${keepOrphans ? "[KEEP ORPHANS] " : ""}Fetching Chutes models from API...`);
+
+  const res = await fetch(API_ENDPOINT);
+  if (!res.ok) {
+    console.error(`Failed to fetch API: ${res.status} ${res.statusText}`);
+    process.exit(1);
+  }
+
+  const json = await res.json();
+  const parsed = ChutesResponse.safeParse(json);
+  if (!parsed.success) {
+    console.error("Invalid API response:", parsed.error.errors);
+    process.exit(1);
+  }
+
+  const apiModels = parsed.data.data;
+
+  // Scan existing TOML files
+  const existingFiles = new Set<string>();
+  try {
+    for await (const file of new Bun.Glob("**/*.toml").scan({
+      cwd: modelsDir,
+      absolute: false,
+    })) {
+      existingFiles.add(file);
+    }
+  } catch {
+  }
+
+  console.log(`Found ${apiModels.length} models in API, ${existingFiles.size} existing files\n`);
+
+  const apiModelIds = new Set<string>();
+
+  let created = 0;
+  let updated = 0;
+  let unchanged = 0;
+
+  for (const apiModel of apiModels) {
+    const relativePath = `${apiModel.id}.toml`;
+    const filePath = path.join(modelsDir, relativePath);
+    const dirPath = path.dirname(filePath);
+
+    apiModelIds.add(relativePath);
+
+    const existing = await loadExistingModel(filePath);
+    const merged = mergeModel(apiModel, existing);
+    const tomlContent = formatToml(merged);
+
+    if (existing === null) {
+      created++;
+      if (dryRun) {
+        console.log(`[DRY RUN] Would create: ${relativePath}`);
+        console.log(`  name = "${merged.name}"`);
+        if (merged.family) {
+          console.log(`  family = "${merged.family}" (inferred)`);
+        }
+        console.log("");
+      } else {
+        await mkdir(dirPath, { recursive: true });
+        await Bun.write(filePath, tomlContent);
+        console.log(`Created: ${relativePath}`);
+      }
+    } else {
+      if (newOnly) {
+        unchanged++;
+        continue;
+      }
+
+      const changes = detectChanges(existing, merged);
+      const existingContent = await Bun.file(filePath).text();
+      const formatChanged = existingContent !== tomlContent;
+
+      if (changes.length > 0 || formatChanged) {
+        updated++;
+        if (dryRun) {
+          console.log(`[DRY RUN] Would update: ${relativePath}`);
+        } else {
+          await mkdir(dirPath, { recursive: true });
+          await Bun.write(filePath, tomlContent);
+          console.log(`Updated: ${relativePath}`);
+        }
+        for (const change of changes) {
+          console.log(`  ${change.field}: ${change.oldValue} → ${change.newValue}`);
+        }
+        if (changes.length === 0 && formatChanged) {
+          console.log(`  (format-only change)`);
+        }
+        console.log("");
+      } else {
+        unchanged++;
+      }
+    }
+  }
+
+  // Handle orphaned files (on disk but not in API)
+  const orphaned: string[] = [];
+  for (const file of existingFiles) {
+    if (!apiModelIds.has(file)) {
+      orphaned.push(file);
+      const orphanPath = path.join(modelsDir, file);
+      if (keepOrphans) {
+        console.log(`Orphaned (kept): ${file}`);
+      } else if (dryRun) {
+        console.log(`[DRY RUN] Would delete: ${file}`);
+      } else {
+        await Bun.file(orphanPath).delete();
+        console.log(`Deleted: ${file}`);
+
+        // Clean up empty parent directories
+        const parentDir = path.dirname(orphanPath);
+        try {
+          const remaining = [];
+          for await (const entry of new Bun.Glob("*").scan({ cwd: parentDir })) {
+            remaining.push(entry);
+          }
+          if (remaining.length === 0) {
+            const { rmdir } = await import("node:fs/promises");
+            await rmdir(parentDir);
+            console.log(`  Removed empty directory: ${path.basename(parentDir)}/`);
+          }
+        } catch {
+          // Directory not empty or other error, ignore
+        }
+      }
+    }
+  }
+
+  console.log("");
+  if (dryRun) {
+    console.log(
+      `Summary: ${created} would be created, ${updated} would be updated, ${unchanged} unchanged, ${orphaned.length} would be deleted`,
+    );
+  } else if (keepOrphans) {
+    console.log(
+      `Summary: ${created} created, ${updated} updated, ${unchanged} unchanged, ${orphaned.length} orphaned (kept)`,
+    );
+  } else {
+    console.log(
+      `Summary: ${created} created, ${updated} updated, ${unchanged} unchanged, ${orphaned.length} deleted`,
+    );
+  }
+}
+
+await main();

packages/core/script/generate-digitalocean.ts

@@ -209,7 +209,18 @@ interface ExistingModel {
       output?: number;
       cache_read?: number;
       cache_write?: number;
+      context_min?: number;
     };
+    tiers?: Array<{
+      tier: {
+        type?: "context";
+        size: number;
+      };
+      input?: number;
+      output?: number;
+      cache_read?: number;
+      cache_write?: number;
+    }>;
   };
   limit?: {
     context?: number;
@@ -262,6 +273,7 @@ interface MergedModel {
       output: number;
       cache_read?: number;
       cache_write?: number;
+      context_min?: number;
     };
   };
   limit: {
@@ -310,6 +322,35 @@ function inferFamily(modelId: string, modelName: string): string | undefined {
   return undefined;
 }
 
+function getExistingLongContextCost(existing: ExistingModel | null) {
+  const tier = existing?.cost?.tiers?.find(
+    (tier) =>
+      (tier.tier.type === undefined || tier.tier.type === "context") &&
+      tier.tier.size >= 200_000,
+  );
+  if (tier) {
+    return {
+      ...tier,
+      context_min: tier.tier.size,
+    };
+  }
+
+  return existing?.cost?.context_over_200k === undefined
+    ? undefined
+    : {
+        ...existing.cost.context_over_200k,
+        context_min: 200_000,
+      };
+}
+
+function getLongContextMin(cost: { context_min?: number }) {
+  return cost.context_min ?? 200_000;
+}
+
+function formatInlineNumber(n: number): string {
+  return n >= 1000 ? n.toString().replace(/\B(?=(\d{3})+(?!\d))/g, "_") : n.toString();
+}
+
 // ---------------------------------------------------------------------------
 // Merge API data with existing TOML
 // ---------------------------------------------------------------------------
@@ -394,27 +435,30 @@ function mergeModel(
     };
 
     // Context-tiered pricing (>200k) from the static-content API
+    const existingLongContextCost = getExistingLongContextCost(existing);
     if (pricing?.inputOver200k !== undefined && pricing?.outputOver200k !== undefined) {
       merged.cost.context_over_200k = {
         input: pricing.inputOver200k,
         output: pricing.outputOver200k,
-        ...(existing?.cost?.context_over_200k?.cache_read !== undefined && {
-          cache_read: existing.cost.context_over_200k.cache_read,
+        context_min: existingLongContextCost?.context_min ?? 200_000,
+        ...(existingLongContextCost?.cache_read !== undefined && {
+          cache_read: existingLongContextCost.cache_read,
         }),
-        ...(existing?.cost?.context_over_200k?.cache_write !== undefined && {
-          cache_write: existing.cost.context_over_200k.cache_write,
+        ...(existingLongContextCost?.cache_write !== undefined && {
+          cache_write: existingLongContextCost.cache_write,
         }),
       };
-    } else if (existing?.cost?.context_over_200k) {
-      // Preserve manually-entered context_over_200k if API has no data
+    } else if (existingLongContextCost) {
+      // Preserve manually-entered tiered pricing if API has no data
       merged.cost.context_over_200k = {
-        input: existing.cost.context_over_200k.input ?? inputPrice,
-        output: existing.cost.context_over_200k.output ?? outputPrice,
-        ...(existing.cost.context_over_200k.cache_read !== undefined && {
-          cache_read: existing.cost.context_over_200k.cache_read,
+        input: existingLongContextCost.input ?? inputPrice,
+        output: existingLongContextCost.output ?? outputPrice,
+        context_min: existingLongContextCost.context_min,
+        ...(existingLongContextCost.cache_read !== undefined && {
+          cache_read: existingLongContextCost.cache_read,
         }),
-        ...(existing.cost.context_over_200k.cache_write !== undefined && {
-          cache_write: existing.cost.context_over_200k.cache_write,
+        ...(existingLongContextCost.cache_write !== undefined && {
+          cache_write: existingLongContextCost.cache_write,
         }),
       };
     }
@@ -463,7 +507,8 @@ function formatToml(model: MergedModel): string {
 
     if (model.cost.context_over_200k) {
       lines.push("");
-      lines.push(`[cost.context_over_200k]`);
+      lines.push(`[[cost.tiers]]`);
+      lines.push(`tier = { size = ${formatInlineNumber(getLongContextMin(model.cost.context_over_200k))} }`);
       lines.push(`input = ${model.cost.context_over_200k.input}`);
       lines.push(`output = ${model.cost.context_over_200k.output}`);
       if (model.cost.context_over_200k.cache_read !== undefined)
@@ -524,8 +569,9 @@ function detectChanges(existing: ExistingModel | null, merged: MergedModel): Cha
   compare("status", existing.status, merged.status);
   compare("cost.input", existing.cost?.input, merged.cost?.input);
   compare("cost.output", existing.cost?.output, merged.cost?.output);
-  compare("cost.context_over_200k.input", existing.cost?.context_over_200k?.input, merged.cost?.context_over_200k?.input);
-  compare("cost.context_over_200k.output", existing.cost?.context_over_200k?.output, merged.cost?.context_over_200k?.output);
+  const existingLongContextCost = getExistingLongContextCost(existing);
+  compare("cost.context_over_200k.input", existingLongContextCost?.input, merged.cost?.context_over_200k?.input);
+  compare("cost.context_over_200k.output", existingLongContextCost?.output, merged.cost?.context_over_200k?.output);
   compare("limit.context", existing.limit?.context, merged.limit.context);
   compare("limit.output", existing.limit?.output, merged.limit.output);
   compare("modalities.input", existing.modalities?.input, merged.modalities.input);

packages/core/script/generate-venice.ts

@@ -162,7 +162,18 @@ interface ExistingModel {
       output?: number;
       cache_read?: number;
       cache_write?: number;
+      context_min?: number;
     };
+    tiers?: Array<{
+      tier: {
+        type?: "context";
+        size: number;
+      };
+      input?: number;
+      output?: number;
+      cache_read?: number;
+      cache_write?: number;
+    }>;
   };
   limit?: {
     context?: number;
@@ -195,6 +206,30 @@ async function loadExistingModel(filePath: string): Promise<ExistingModel | null
   }
 }
 
+function getExistingLongContextMin(existing: ExistingModel | null) {
+  return (
+    existing?.cost?.tiers?.find(
+      (tier) =>
+        (tier.tier.type === undefined || tier.tier.type === "context") &&
+        tier.tier.size >= 200_000,
+    )?.tier.size ?? 200_000
+  );
+}
+
+function getExistingLongContextCost(existing: ExistingModel | null) {
+  return (
+    existing?.cost?.tiers?.find(
+      (tier) =>
+        (tier.tier.type === undefined || tier.tier.type === "context") &&
+        tier.tier.size >= 200_000,
+    ) ?? existing?.cost?.context_over_200k
+  );
+}
+
+function getLongContextMin(cost: { context_min?: number }) {
+  return cost.context_min ?? 200_000;
+}
+
 interface MergedModel {
   name: string;
   family?: string;
@@ -219,6 +254,7 @@ interface MergedModel {
       output: number;
       cache_read?: number;
       cache_write?: number;
+      context_min?: number;
     };
   };
   limit: {
@@ -292,6 +328,7 @@ function mergeModel(
       merged.cost.context_over_200k = {
         input: spec.pricing.extended.input.usd,
         output: spec.pricing.extended.output.usd,
+        context_min: spec.pricing.extended.context_token_threshold,
         ...(spec.pricing.extended.cache_input && { cache_read: spec.pricing.extended.cache_input.usd }),
         ...(spec.pricing.extended.cache_write && { cache_write: spec.pricing.extended.cache_write.usd }),
       };
@@ -366,7 +403,8 @@ function formatToml(model: MergedModel): string {
 
     if (model.cost.context_over_200k) {
       lines.push("");
-      lines.push(`[cost.context_over_200k]`);
+      lines.push(`[[cost.tiers]]`);
+      lines.push(`tier = { size = ${formatNumber(getLongContextMin(model.cost.context_over_200k))} }`);
       lines.push(`input = ${model.cost.context_over_200k.input}`);
       lines.push(`output = ${model.cost.context_over_200k.output}`);
       if (model.cost.context_over_200k.cache_read !== undefined) {
@@ -438,10 +476,11 @@ function detectChanges(
   compare("cost.output", existing.cost?.output, merged.cost?.output);
   compare("cost.cache_read", existing.cost?.cache_read, merged.cost?.cache_read);
   compare("cost.cache_write", existing.cost?.cache_write, merged.cost?.cache_write);
-  compare("cost.context_over_200k.input", existing.cost?.context_over_200k?.input, merged.cost?.context_over_200k?.input);
-  compare("cost.context_over_200k.output", existing.cost?.context_over_200k?.output, merged.cost?.context_over_200k?.output);
-  compare("cost.context_over_200k.cache_read", existing.cost?.context_over_200k?.cache_read, merged.cost?.context_over_200k?.cache_read);
-  compare("cost.context_over_200k.cache_write", existing.cost?.context_over_200k?.cache_write, merged.cost?.context_over_200k?.cache_write);
+  const existingLongContextCost = getExistingLongContextCost(existing);
+  compare("cost.context_over_200k.input", existingLongContextCost?.input, merged.cost?.context_over_200k?.input);
+  compare("cost.context_over_200k.output", existingLongContextCost?.output, merged.cost?.context_over_200k?.output);
+  compare("cost.context_over_200k.cache_read", existingLongContextCost?.cache_read, merged.cost?.context_over_200k?.cache_read);
+  compare("cost.context_over_200k.cache_write", existingLongContextCost?.cache_write, merged.cost?.context_over_200k?.cache_write);
   compare("limit.context", existing.limit?.context, merged.limit.context);
   compare("limit.output", existing.limit?.output, merged.limit.output);
   compare("modalities.input", existing.modalities?.input, merged.modalities.input);

packages/core/src/family.ts

@@ -55,6 +55,7 @@ export const ModelFamilyValues = [
   "deepseek",
   "deepseek-thinking",
   "deepseek-flash",
+  "deepseek-flash-free",
   "deepseek-flash-think",
 
   // Microsoft Phi
@@ -289,12 +290,14 @@ export const ModelFamilyValues = [
   "rnj",
 
   // Tecent Hy
+  "hy3",
   "hy3-free",
 
   // Ling & Ring (InclusionAI)
   "ling",
   "ling-flash-free",
   "ring",
+  "ring-1t-free",
 
   // Kat Coder
   "kat-coder",

packages/core/src/generate.ts

@@ -2,9 +2,9 @@ import path from "path";
 import { mergeDeep } from "remeda";
 import { z } from "zod";
 
-import { Provider, Model } from "./schema.js";
+import { Provider, Model, AuthoredModel, AuthoredModelShape } from "./schema.js";
 
-const ExtendsModel = Model.sourceType()
+const ExtendsModel = AuthoredModelShape
   .partial()
   .extend({
     extends: z
@@ -71,12 +71,12 @@ export async function generate(directory: string) {
         });
         continue;
       }
-      const model = Model.safeParse(toml);
+      const model = AuthoredModel.safeParse(toml);
       if (!model.success) {
         model.error.cause = { modelPath, toml };
         throw model.error;
       }
-      provider.data.models[modelID] = model.data;
+      provider.data.models[modelID] = normalizeModelCost(model.data);
     }
     result[providerID] = provider.data;
   }
@@ -144,7 +144,7 @@ export async function generate(directory: string) {
       }
     }
 
-    const model = Model.safeParse(merged);
+    const model = Model.safeParse(normalizeCost(merged));
     if (!model.success) {
       model.error.cause = { modelPath: pendingModel.modelPath, toml: merged };
       throw model.error;
@@ -155,3 +155,51 @@ export async function generate(directory: string) {
 
   return result;
 }
+
+function normalizeModelCost(model: z.infer<typeof AuthoredModel>): Model {
+  return normalizeCost(model) as Model;
+}
+
+function normalizeCost(model: Record<string, unknown>) {
+  const cost = model.cost;
+  if (cost === undefined || cost === null || typeof cost !== "object" || Array.isArray(cost)) {
+    return model;
+  }
+
+  const tiers = (cost as { tiers?: unknown }).tiers;
+  if (!Array.isArray(tiers)) {
+    return model;
+  }
+
+  if (tiers.length !== 1) {
+    return model;
+  }
+
+  const contextOver200k = tiers.find((tier) => {
+    if (tier === null || typeof tier !== "object" || Array.isArray(tier)) return false;
+    const tierConfig = (tier as { tier?: unknown }).tier;
+    if (tierConfig === null || typeof tierConfig !== "object" || Array.isArray(tierConfig)) return false;
+    const type = (tierConfig as { type?: unknown }).type;
+    const size = (tierConfig as { size?: unknown }).size;
+    // context_over_200k is a legacy compatibility field. It intentionally
+    // includes higher thresholds; cost.tiers carries the exact threshold.
+    return (
+      (type === undefined || type === "context") &&
+      typeof size === "number" &&
+      size >= 200_000
+    );
+  });
+
+  if (contextOver200k === undefined) {
+    return model;
+  }
+
+  const { tier: _tier, ...legacyCost } = contextOver200k as Record<string, unknown>;
+  return {
+    ...model,
+    cost: {
+      ...(cost as Record<string, unknown>),
+      context_over_200k: legacyCost,
+    },
+  };
+}

packages/core/src/schema.ts

@@ -21,110 +21,164 @@ const JsonValue: z.ZodType<JsonValue> = z.lazy(() =>
   ]),
 );
 
-const Cost = z.object({
-  input: z.number().min(0, "Input price cannot be negative"),
-  output: z.number().min(0, "Output price cannot be negative"),
-  reasoning: z.number().min(0, "Input price cannot be negative").optional(),
-  cache_read: z
-    .number()
-    .min(0, "Cache read price cannot be negative")
+const Cost = z
+  .object({
+    input: z.number().min(0, "Input price cannot be negative"),
+    output: z.number().min(0, "Output price cannot be negative"),
+    reasoning: z
+      .number()
+      .min(0, "Reasoning price cannot be negative")
+      .optional(),
+    cache_read: z
+      .number()
+      .min(0, "Cache read price cannot be negative")
+      .optional(),
+    cache_write: z
+      .number()
+      .min(0, "Cache write price cannot be negative")
+      .optional(),
+    input_audio: z
+      .number()
+      .min(0, "Audio input price cannot be negative")
+      .optional(),
+    output_audio: z
+      .number()
+      .min(0, "Audio output price cannot be negative")
+      .optional(),
+  });
+
+const CostTier = Cost.extend({
+  tier: z
+    .object({
+      type: z.literal("context").default("context"),
+      size: z.number().int().min(0, "Context tier size cannot be negative"),
+    })
+    .strict(),
+}).strict();
+
+const AuthoredCost = Cost.extend({
+  context_over_200k: z.never().optional(),
+  tiers: z.array(CostTier).optional(),
+});
+
+const OutputCost = Cost.extend({
+  context_over_200k: Cost.optional(),
+  tiers: z.array(CostTier).optional(),
+});
+
+const ModelBase = z.object({
+  id: z.string(),
+  name: z.string().min(1, "Model name cannot be empty"),
+  family: ModelFamily.optional(),
+  attachment: z.boolean(),
+  reasoning: z.boolean(),
+  tool_call: z.boolean(),
+  interleaved: z
+    .union([
+      z.literal(true),
+      z
+        .object({
+          field: z.enum(["reasoning_content", "reasoning_details"]),
+        })
+        .strict(),
+    ])
     .optional(),
-  cache_write: z
-    .number()
-    .min(0, "Cache write price cannot be negative")
+  structured_output: z.boolean().optional(),
+  temperature: z.boolean().optional(),
+  knowledge: z
+    .string()
+    .regex(/^\d{4}-\d{2}(-\d{2})?$/, {
+      message: "Must be in YYYY-MM or YYYY-MM-DD format",
+    })
     .optional(),
-  input_audio: z
-    .number()
-    .min(0, "Audio input price cannot be negative")
+  release_date: z.string().regex(/^\d{4}-\d{2}(-\d{2})?$/, {
+    message: "Must be in YYYY-MM or YYYY-MM-DD format",
+  }),
+  last_updated: z.string().regex(/^\d{4}-\d{2}(-\d{2})?$/, {
+    message: "Must be in YYYY-MM or YYYY-MM-DD format",
+  }),
+  modalities: z.object({
+    input: z.array(z.enum(["text", "audio", "image", "video", "pdf"])),
+    output: z.array(z.enum(["text", "audio", "image", "video", "pdf"])),
+  }),
+  open_weights: z.boolean(),
+  limit: z.object({
+    context: z.number().min(0, "Context window must be positive"),
+    input: z.number().min(0, "Input tokens must be positive").optional(),
+    output: z.number().min(0, "Output tokens must be positive"),
+  }),
+  status: z.enum(["alpha", "beta", "deprecated"]).optional(),
+  experimental: z
+    .object({
+      modes: z
+        .record(
+          z.object({
+            cost: Cost.optional(),
+            provider: z
+              .object({
+                body: z.record(JsonValue).optional(),
+                headers: z.record(z.string()).optional(),
+              })
+              .optional(),
+          }),
+        )
+        .optional(),
+    })
     .optional(),
-  output_audio: z
-    .number()
-    .min(0, "Audio output price cannot be negative")
+  provider: z
+    .object({
+      npm: z.string().optional(),
+      api: z.string().optional(),
+      shape: z.enum(["responses", "completions"]).optional(),
+      body: z.record(JsonValue).optional(),
+      headers: z.record(z.string()).optional(),
+    })
     .optional(),
 });
-export const Model = z
+
+function refineModel<T extends z.ZodTypeAny>(schema: T) {
+  return schema
+    .refine(
+      (data) => {
+        return !(data.reasoning === false && data.cost?.reasoning !== undefined);
+      },
+      {
+        message: "Cannot set cost.reasoning when reasoning is false",
+        path: ["cost", "reasoning"],
+      },
+    )
+    .refine(
+      (data) => {
+        const tiers = data.cost?.tiers;
+        if (tiers === undefined) return true;
+
+        const sizes = tiers.map((tier: { tier: { size: number } }) => tier.tier.size);
+        return new Set(sizes).size === sizes.length;
+      },
+      {
+        message: "Cost context tiers must not have duplicate sizes",
+        path: ["cost", "tiers"],
+      },
+    );
+}
+
+export const ModelShape = z
   .object({
-    id: z.string(),
-    name: z.string().min(1, "Model name cannot be empty"),
-    family: ModelFamily.optional(),
-    attachment: z.boolean(),
-    reasoning: z.boolean(),
-    tool_call: z.boolean(),
-    interleaved: z
-      .union([
-        z.literal(true),
-        z
-          .object({
-            field: z.enum(["reasoning_content", "reasoning_details"]),
-          })
-          .strict(),
-      ])
-      .optional(),
-    structured_output: z.boolean().optional(),
-    temperature: z.boolean().optional(),
-    knowledge: z
-      .string()
-      .regex(/^\d{4}-\d{2}(-\d{2})?$/, {
-        message: "Must be in YYYY-MM or YYYY-MM-DD format",
-      })
-      .optional(),
-    release_date: z.string().regex(/^\d{4}-\d{2}(-\d{2})?$/, {
-      message: "Must be in YYYY-MM or YYYY-MM-DD format",
-    }),
-    last_updated: z.string().regex(/^\d{4}-\d{2}(-\d{2})?$/, {
-      message: "Must be in YYYY-MM or YYYY-MM-DD format",
-    }),
-    modalities: z.object({
-      input: z.array(z.enum(["text", "audio", "image", "video", "pdf"])),
-      output: z.array(z.enum(["text", "audio", "image", "video", "pdf"])),
-    }),
-    open_weights: z.boolean(),
-    cost: Cost.extend({
-      context_over_200k: Cost.optional(),
-    }).optional(),
-    limit: z.object({
-      context: z.number().min(0, "Context window must be positive"),
-      input: z.number().min(0, "Input tokens must be positive").optional(),
-      output: z.number().min(0, "Output tokens must be positive"),
-    }),
-    status: z.enum(["alpha", "beta", "deprecated"]).optional(),
-    experimental: z
-      .object({
-        modes: z
-          .record(
-            z.object({
-              cost: Cost.optional(),
-              provider: z
-                .object({
-                  body: z.record(JsonValue).optional(),
-                  headers: z.record(z.string()).optional(),
-                })
-                .optional(),
-            }),
-          )
-          .optional(),
-      })
-      .optional(),
-    provider: z
-      .object({
-        npm: z.string().optional(),
-        api: z.string().optional(),
-        shape: z.enum(["responses", "completions"]).optional(),
-        body: z.record(JsonValue).optional(),
-        headers: z.record(z.string()).optional(),
-      })
-      .optional(),
+    ...ModelBase.shape,
+    cost: OutputCost.optional(),
   })
-  .strict()
-  .refine(
-    (data) => {
-      return !(data.reasoning === false && data.cost?.reasoning !== undefined);
-    },
-    {
-      message: "Cannot set cost.reasoning when reasoning is false",
-      path: ["cost", "reasoning"],
-    },
-  );
+  .strict();
+
+export const AuthoredModelShape = z
+  .object({
+    ...ModelBase.shape,
+    cost: AuthoredCost.optional(),
+  })
+  .strict();
+
+export const Model = refineModel(ModelShape);
+
+export const AuthoredModel = refineModel(AuthoredModelShape);
 
 export type Model = z.infer<typeof Model>;
 

packages/web/package.json

@@ -6,7 +6,6 @@
     "build": "./script/build.ts"
   },
   "dependencies": {
-    "@tanstack/virtual-core": "^3.14.0",
     "hono": "^4.8.0",
     "models.dev": "workspace:*"
   },

packages/web/src/index.css

@@ -41,8 +41,6 @@
 
 html,
 body {
-  height: 100%;
-  overflow: hidden;
   font-family: 'Rubik', sans-serif;
   line-height: 1.6;
   color: var(--color-text);
@@ -207,17 +205,12 @@ header {
   }
 }
 
-.table-viewport {
-  height: calc(100svh - var(--header-height));
-  margin-top: var(--header-height);
-  overflow: auto;
-}
-
 table {
   border-collapse: separate;
   border-spacing: 0;
   font-size: 0.875rem;
   width: 100%;
+  margin-top: var(--header-height);
 }
 
 thead,
@@ -225,7 +218,7 @@ tbody {}
 
 table thead th {
   position: sticky;
-  top: 0;
+  top: var(--header-height);
   border-top: 1px solid var(--color-border);
   border-bottom: 1px solid var(--color-border);
   font-size: 0.75rem;
@@ -271,7 +264,6 @@ td {
   text-align: left;
   border-bottom: 1px solid var(--color-border);
   white-space: nowrap;
-  height: 48px;
 }
 
 tbody {
@@ -327,37 +319,21 @@ tbody {
     gap: 0.375rem;
   }
 
-  .provider-cell img,
-  .provider-cell svg {
+  .provider-cell span:first-child {
     flex: 0 0 auto;
+  }
+
+  .provider-cell svg {
     display: block;
     width: 1rem;
     height: 1rem;
     color: var(--color-text-secondary);
   }
 
-  .virtual-spacer td {
-    height: inherit;
-    padding: 0;
-    border-bottom: 0;
-  }
-
-  .empty-row td {
-    padding: 2rem 0.75rem;
-    color: var(--color-text-tertiary);
-  }
-
-  .empty-row div {
-    position: sticky;
-    left: 0;
-    width: calc(100vw - 1.5rem);
-    text-align: center;
-  }
-
   .model-id-cell {
     display: flex;
     align-items: center;
-    justify-content: flex-start;
+    justify-content: space-between;
     gap: 0.375rem;
   }
 
@@ -570,4 +546,4 @@ dialog {
     }
   }
 
-}
+}

packages/web/src/index.ts

@@ -1,53 +1,7 @@
-import {
-  Virtualizer,
-  elementScroll,
-  observeElementOffset,
-  observeElementRect,
-} from "@tanstack/virtual-core";
-import {
-  type TableRow,
-  renderRow,
-  escapeHtml,
-  booleanText,
-  knowledgeText,
-  weightsText,
-} from "./shared.js";
-
-declare global {
-  interface Window {
-    __TABLE_DATA__: TableRow[];
-  }
-}
-
-interface VirtualizedRow extends TableRow {
-  key: string;
-  searchText: string;
-  sortValues: Array<string | number | undefined>;
-}
-
-type SortDirection = "asc" | "desc";
-
-const ESTIMATED_ROW_HEIGHT = 48;
-const VIRTUAL_OVERSCAN = 5;
-
 const modal = document.getElementById("modal") as HTMLDialogElement;
 const modalClose = document.getElementById("close")!;
 const help = document.getElementById("help")!;
 const search = document.getElementById("search")! as HTMLInputElement;
-const viewport = document.getElementById("table-viewport") as HTMLElement;
-const tbody = document.getElementById(
-  "models-table-body"
-) as HTMLTableSectionElement;
-const headers = Array.from(document.querySelectorAll("th.sortable"));
-const columnCount = document.querySelectorAll("thead th").length;
-
-let isLoaded = false;
-let allRows: VirtualizedRow[] = [];
-let visibleRows: VirtualizedRow[] = [];
-let currentSort: { column: number; direction: SortDirection } = {
-  column: -1,
-  direction: "asc",
-};
 
 /////////////////////////
 // URL State Management
@@ -77,7 +31,10 @@ function getColumnNameForURL(headerEl: Element): string {
 }
 
 function getColumnIndexByUrlName(name: string): number {
-  return headers.findIndex((header) => getColumnNameForURL(header) === name);
+  const headers = document.querySelectorAll("th.sortable");
+  return Array.from(headers).findIndex(
+    (header) => getColumnNameForURL(header) === name
+  );
 }
 
 /////////////////////////
@@ -106,181 +63,32 @@ modal.addEventListener("click", (e) => {
 });
 
 ////////////////////
-// Row Data
+// Handle Sorting
 ////////////////////
-function lockColumnWidths() {
-  const ths = document.querySelectorAll("#models-table thead th");
-  const widths = Array.from(ths).map((th) => th.getBoundingClientRect().width);
+let currentSort = { column: -1, direction: "asc" };
 
-  const measurementRow = tbody.querySelector('tr[aria-hidden="true"]');
-  if (measurementRow) measurementRow.remove();
+function sortTable(column: number, direction: "asc" | "desc") {
+  const header = document.querySelectorAll("th.sortable")[column];
+  const columnType = header.getAttribute("data-type");
+  if (!columnType) return;
 
-  const table = document.getElementById("models-table")!;
-  table.style.tableLayout = "fixed";
+  // update state
+  currentSort = { column, direction };
+  updateQueryParams({
+    sort: getColumnNameForURL(header),
+    order: direction,
+  });
 
-  const colgroup = document.createElement("colgroup");
-  for (const width of widths) {
-    const col = document.createElement("col");
-    col.style.width = `${width}px`;
-    colgroup.appendChild(col);
-  }
-  table.insertBefore(colgroup, table.firstChild);
-}
-
-function prepareRow(row: TableRow): VirtualizedRow {
-  const sortValues: VirtualizedRow["sortValues"] = [
-    row.providerName,
-    row.modelName,
-    row.family,
-    row.providerId,
-    row.modelId,
-    booleanText(row.toolCall),
-    booleanText(row.reasoning),
-    row.input.length,
-    row.output.length,
-    row.inputCost,
-    row.outputCost,
-    row.reasoningCost,
-    row.cacheReadCost,
-    row.cacheWriteCost,
-    row.audioInputCost,
-    row.audioOutputCost,
-    row.contextLimit,
-    row.inputLimit,
-    row.outputLimit,
-    row.structuredOutput === undefined
-      ? undefined
-      : booleanText(row.structuredOutput),
-    booleanText(row.temperature),
-    weightsText(row.openWeights),
-    row.knowledge ? knowledgeText(row.knowledge) : undefined,
-    row.releaseDate,
-    row.lastUpdated,
-  ];
-
-  const searchableValues = [
-    row.providerName,
-    row.modelName,
-    row.family ?? "",
-    row.providerId,
-    row.modelId,
-    row.releaseDate,
-    row.lastUpdated,
-  ];
-
-  return {
-    ...row,
-    key: `${row.providerId}/${row.modelId}`,
-    searchText: searchableValues.join(" ").toLowerCase(),
-    sortValues,
-  };
-}
-
-////////////////////
-// Virtual Table
-////////////////////
-function getVirtualizerOptions(count: number) {
-  return {
-    count,
-    getScrollElement: () => viewport,
-    estimateSize: () => ESTIMATED_ROW_HEIGHT,
-    getItemKey: (index: number) => visibleRows[index]?.key ?? index,
-    initialRect: {
-      width: viewport.clientWidth || window.innerWidth,
-      height: viewport.clientHeight || window.innerHeight,
-    },
-    overscan: VIRTUAL_OVERSCAN,
-    observeElementRect,
-    observeElementOffset,
-    scrollToFn: elementScroll,
-    onChange: () => renderVirtualRows(),
-  };
-}
-
-const virtualizer = new Virtualizer<HTMLElement, HTMLTableRowElement>(
-  getVirtualizerOptions(0)
-);
-const cleanupVirtualizer = virtualizer._didMount();
-virtualizer._willUpdate();
-window.addEventListener("pagehide", () => cleanupVirtualizer());
-
-function renderStatusRow(message: string) {
-  tbody.innerHTML = `<tr class="empty-row"><td colspan="${columnCount}"><div>${escapeHtml(
-    message
-  )}</div></td></tr>`;
-}
-
-function setVirtualizerCount(count: number, resetScroll: boolean) {
-  virtualizer.setOptions(getVirtualizerOptions(count));
-  virtualizer._willUpdate();
-  if (resetScroll) virtualizer.scrollToOffset(0);
-  renderVirtualRows();
-}
-
-function renderVirtualRows() {
-  if (!isLoaded) return;
-  if (visibleRows.length === 0) {
-    renderStatusRow("No models found");
-    return;
-  }
-
-  const virtualRows = virtualizer.getVirtualItems();
-  if (virtualRows.length === 0) return;
-
-  const firstRow = virtualRows[0]!;
-  const lastRow = virtualRows[virtualRows.length - 1]!;
-  const paddingTop = firstRow.start;
-  const paddingBottom = Math.max(virtualizer.getTotalSize() - lastRow.end, 0);
-  const html: string[] = [];
-
-  if (paddingTop > 0) html.push(renderSpacerRow(paddingTop));
-  for (const virtualRow of virtualRows) {
-    const row = visibleRows[virtualRow.index];
-    if (row) html.push(renderRow(row, virtualRow.index));
-  }
-  if (paddingBottom > 0) html.push(renderSpacerRow(paddingBottom));
-
-  tbody.innerHTML = html.join("");
-  tbody.querySelectorAll<HTMLTableRowElement>("tr[data-index]").forEach((row) =>
-    virtualizer.measureElement(row)
-  );
-}
-
-function renderSpacerRow(height: number) {
-  return `<tr class="virtual-spacer" style="height: ${height}px"><td colspan="${columnCount}"></td></tr>`;
-}
-
-function applyRows(resetScroll = true) {
-  if (!isLoaded) return;
-  visibleRows = getRowsForDisplay();
-  setVirtualizerCount(visibleRows.length, resetScroll);
-}
-
-////////////////////
-// Sorting
-////////////////////
-function getRowsForDisplay() {
-  const terms = search.value
-    .toLowerCase()
-    .split(",")
-    .map((term) => term.trim())
-    .filter(Boolean);
-  const filteredRows =
-    terms.length === 0
-      ? allRows
-      : allRows.filter((row) =>
-          terms.some((term) => row.searchText.includes(term))
-        );
-
-  if (currentSort.column === -1) return filteredRows;
-
-  const columnType = headers[currentSort.column]?.getAttribute("data-type");
-  if (!columnType) return filteredRows;
-
-  return [...filteredRows].sort((a, b) => {
-    const aValue = a.sortValues[currentSort.column];
-    const bValue = b.sortValues[currentSort.column];
+  // sort rows
+  const tbody = document.querySelector("table tbody")!;
+  const rows = Array.from(
+    tbody.querySelectorAll("tr")
+  ) as HTMLTableRowElement[];
+  rows.sort((a, b) => {
+    const aValue = getCellValue(a.cells[column], columnType);
+    const bValue = getCellValue(b.cells[column], columnType);
 
+    // Handle undefined values - always sort to bottom
     if (aValue === undefined && bValue === undefined) return 0;
     if (aValue === undefined) return 1;
     if (bValue === undefined) return -1;
@@ -288,48 +96,45 @@ function getRowsForDisplay() {
     let comparison = 0;
     if (columnType === "number" || columnType === "modalities") {
       comparison = (aValue as number) - (bValue as number);
+    } else if (columnType === "boolean") {
+      comparison = (aValue as string).localeCompare(bValue as string);
     } else {
-      comparison = String(aValue).localeCompare(String(bValue));
+      comparison = (aValue as string).localeCompare(bValue as string);
     }
 
-    return currentSort.direction === "asc" ? comparison : -comparison;
+    return direction === "asc" ? comparison : -comparison;
   });
-}
+  rows.forEach((row) => tbody.appendChild(row));
 
-function sortTable(
-  column: number,
-  direction: SortDirection,
-  updateURL = true
-) {
-  const header = headers[column];
-  if (!header?.getAttribute("data-type")) return;
-
-  currentSort = { column, direction };
-  if (updateURL) {
-    updateQueryParams({
-      sort: getColumnNameForURL(header),
-      order: direction,
-    });
-  }
-
-  updateSortIndicators();
-  applyRows();
-}
-
-function updateSortIndicators() {
+  // update sort indicators
+  const headers = document.querySelectorAll("th.sortable");
   headers.forEach((header, i) => {
     const indicator = header.querySelector(".sort-indicator")!;
-    indicator.textContent =
-      i === currentSort.column
-        ? currentSort.direction === "asc"
-          ? "↑"
-          : "↓"
-        : "";
+
+    if (i === column) {
+      indicator.textContent = direction === "asc" ? "↑" : "↓";
+    } else {
+      indicator.textContent = "";
+    }
   });
 }
 
-headers.forEach((header, column) => {
+function getCellValue(
+  cell: HTMLTableCellElement,
+  type: string
+): string | number | undefined {
+  if (type === "modalities")
+    return cell.querySelectorAll(".modality-icon").length;
+
+  const text = cell.textContent?.trim() || "";
+  if (text === "-") return;
+  if (type === "number") return parseFloat(text.replace(/[$,]/g, "")) || 0;
+  return text;
+}
+
+document.querySelectorAll("th.sortable").forEach((header) => {
   header.addEventListener("click", () => {
+    const column = Array.from(header.parentElement!.children).indexOf(header);
     const direction =
       currentSort.column === column && currentSort.direction === "asc"
         ? "desc"
@@ -339,19 +144,34 @@ headers.forEach((header, column) => {
 });
 
 ///////////////////
-// Search
+// Handle Search
 ///////////////////
+function filterTable(value: string) {
+  const lowerCaseValues = value.toLowerCase().split(",").filter(str => str.trim() !== "");
+  const rows = document.querySelectorAll(
+    "table tbody tr"
+  ) as NodeListOf<HTMLTableRowElement>;
+
+  rows.forEach((row) => {
+    const cellTexts = Array.from(row.cells).map((cell) =>
+      cell.textContent!.toLowerCase()
+    );
+    const isVisible = lowerCaseValues.length === 0 ||
+     lowerCaseValues.some((lowerCaseValue) => cellTexts.some((text) => text.includes(lowerCaseValue)));
+    row.style.display = isVisible ? "" : "none";
+  });
+
+  updateQueryParams({ search: value || null });
+}
+
 search.addEventListener("input", () => {
-  updateQueryParams({ search: search.value || null });
-  applyRows();
+  filterTable(search.value);
 });
 
 document.addEventListener("keydown", (e) => {
-  const key = e.key.toLowerCase();
-  if ((e.metaKey || e.ctrlKey) && (key === "k" || key === "f")) {
+  if ((e.metaKey || e.ctrlKey) && e.key === "k") {
     e.preventDefault();
     search.focus();
-    search.select();
   }
 });
 
@@ -365,7 +185,10 @@ search.addEventListener("keydown", (e) => {
 ///////////////////////////////////
 // Handle Copy model ID function
 ///////////////////////////////////
-async function copyModelId(button: HTMLButtonElement, modelId: string) {
+(window as any).copyModelId = async (
+  button: HTMLButtonElement,
+  modelId: string
+) => {
   try {
     if (navigator.clipboard) {
       await navigator.clipboard.writeText(modelId);
@@ -386,18 +209,7 @@ async function copyModelId(button: HTMLButtonElement, modelId: string) {
   } catch (err) {
     console.error("Failed to copy text: ", err);
   }
-}
-
-document.addEventListener("click", (event) => {
-  if (!(event.target instanceof Element)) return;
-  const button = event.target.closest<HTMLButtonElement>(
-    ".copy-button[data-model-id]"
-  );
-  if (!button) return;
-
-  const modelId = button.dataset.modelId;
-  if (modelId) void copyModelId(button, modelId);
-});
+};
 
 ///////////////////////////////////
 // Initialize State from URL
@@ -405,37 +217,24 @@ document.addEventListener("click", (event) => {
 function initializeFromURL() {
   const params = getQueryParams();
 
-  search.value = params.get("search") ?? "";
+  (() => {
+    const searchQuery = params.get("search");
+    if (!searchQuery) return;
+    search.value = searchQuery;
+    filterTable(searchQuery);
+  })();
+
+  (() => {
+    const columnName = params.get("sort");
+    if (!columnName) return;
 
-  currentSort = { column: -1, direction: "asc" };
-  const columnName = params.get("sort");
-  if (columnName) {
     const columnIndex = getColumnIndexByUrlName(columnName);
-    if (columnIndex !== -1) {
-      currentSort = {
-        column: columnIndex,
-        direction: params.get("order") === "desc" ? "desc" : "asc",
-      };
-    }
-  }
+    if (columnIndex === -1) return;
 
-  updateSortIndicators();
-  applyRows(false);
+    const direction = (params.get("order") as "asc" | "desc") || "asc";
+    sortTable(columnIndex, direction);
+  })();
 }
 
-function loadRows() {
-  try {
-    allRows = window.__TABLE_DATA__.map(prepareRow);
-    lockColumnWidths();
-    isLoaded = true;
-    initializeFromURL();
-  } catch (error) {
-    console.error(error);
-    isLoaded = true;
-    visibleRows = [];
-    renderStatusRow("Failed to load model data");
-  }
-}
-
-loadRows();
+document.addEventListener("DOMContentLoaded", initializeFromURL);
 window.addEventListener("popstate", initializeFromURL);

packages/web/src/render.tsx

@@ -6,7 +6,6 @@ import { Fragment } from "hono/jsx";
 import { renderToString } from "hono/jsx/dom/server";
 import { existsSync } from "fs";
 import path from "path";
-import { type TableRow, renderRow, getLargestRow } from "./shared.js";
 
 export const Providers = await generate(
   path.join(import.meta.dir, "..", "..", "..", "providers")
@@ -39,6 +38,7 @@ const loadProviderSvg = async (providerId: string): Promise<string | null> => {
       const file = Bun.file(providerLogoPath);
       return await file.text();
     }
+    //
     // Fall back to default logo
     if (existsSync(defaultLogoPath)) {
       const file = Bun.file(defaultLogoPath);
@@ -62,47 +62,122 @@ for (const [providerId] of Object.entries(Providers)) {
   }
 }
 
-export const INITIAL_ROW_COUNT = 50;
+function renderProviderLogo(providerId: string) {
+  const svgContent = providerLogos.get(providerId) || "";
 
-export const TableRows: TableRow[] = Object.entries(Providers)
-  .sort(([, providerA], [, providerB]) =>
-    providerA.name.localeCompare(providerB.name)
-  )
-  .flatMap(([providerId, provider]) =>
-    Object.entries(provider.models)
-      .filter(([, model]) => model.status !== "alpha")
-      .sort(([, modelA], [, modelB]) => modelA.name.localeCompare(modelB.name))
-      .map(([modelId, model]) => ({
-        providerId,
-        providerName: provider.name,
-        providerLogoSvg: providerLogos.get(providerId) || "",
-        modelId,
-        modelName: model.name,
-        family: model.family,
-        toolCall: model.tool_call,
-        reasoning: model.reasoning,
-        input: model.modalities.input,
-        output: model.modalities.output,
-        inputCost: model.cost?.input,
-        outputCost: model.cost?.output,
-        reasoningCost: model.cost?.reasoning,
-        cacheReadCost: model.cost?.cache_read,
-        cacheWriteCost: model.cost?.cache_write,
-        audioInputCost: model.cost?.input_audio,
-        audioOutputCost: model.cost?.output_audio,
-        contextLimit: model.limit.context,
-        inputLimit: model.limit.input,
-        outputLimit: model.limit.output,
-        structuredOutput: model.structured_output,
-        temperature: model.temperature ?? false,
-        openWeights: model.open_weights,
-        knowledge: model.knowledge,
-        releaseDate: model.release_date,
-        lastUpdated: model.last_updated,
-      }))
-  );
+  return <span dangerouslySetInnerHTML={{ __html: svgContent }} />;
+}
 
-const largestRow = getLargestRow(TableRows);
+const getModalityIcon = (modality: string) => {
+  switch (modality) {
+    case "text":
+      return (
+        <span class="modality-icon" data-tooltip="Text">
+          <svg
+            xmlns="http://www.w3.org/2000/svg"
+            width="16"
+            height="16"
+            viewBox="0 0 24 24"
+            fill="none"
+            stroke="currentColor"
+            stroke-width="2"
+            stroke-linecap="round"
+            stroke-linejoin="round"
+          >
+            <polyline points="4,7 4,4 20,4 20,7"></polyline>
+            <line x1="9" y1="20" x2="15" y2="20"></line>
+            <line x1="12" y1="4" x2="12" y2="20"></line>
+          </svg>
+        </span>
+      );
+    case "image":
+      return (
+        <span class="modality-icon" data-tooltip="Image">
+          <svg
+            xmlns="http://www.w3.org/2000/svg"
+            width="16"
+            height="16"
+            viewBox="0 0 24 24"
+            fill="none"
+            stroke="currentColor"
+            stroke-width="2"
+            stroke-linecap="round"
+            stroke-linejoin="round"
+          >
+            <rect width="18" height="18" x="3" y="3" rx="2" ry="2"></rect>
+            <circle cx="9" cy="9" r="2"></circle>
+            <path d="m21 15-3.086-3.086a2 2 0 0 0-2.828 0L6 21"></path>
+          </svg>
+        </span>
+      );
+    case "audio":
+      return (
+        <span class="modality-icon" data-tooltip="Audio">
+          <svg
+            xmlns="http://www.w3.org/2000/svg"
+            width="16"
+            height="16"
+            viewBox="0 0 24 24"
+            fill="none"
+            stroke="currentColor"
+            stroke-width="2"
+            stroke-linecap="round"
+            stroke-linejoin="round"
+          >
+            <polygon points="11 5 6 9 2 9 2 15 6 15 11 19 11 5"></polygon>
+            <path d="m19.07 4.93a10 10 0 0 1 0 14.14M15.54 8.46a5 5 0 0 1 0 7.07"></path>
+          </svg>
+        </span>
+      );
+    case "video":
+      return (
+        <span class="modality-icon" data-tooltip="Video">
+          <svg
+            xmlns="http://www.w3.org/2000/svg"
+            width="16"
+            height="16"
+            viewBox="0 0 24 24"
+            fill="none"
+            stroke="currentColor"
+            stroke-width="2"
+            stroke-linecap="round"
+            stroke-linejoin="round"
+          >
+            <path d="m22 8-6 4 6 4V8Z"></path>
+            <rect width="14" height="12" x="2" y="6" rx="2" ry="2"></rect>
+          </svg>
+        </span>
+      );
+    case "pdf":
+      return (
+        <span class="modality-icon" data-tooltip="PDF">
+          <svg
+            xmlns="http://www.w3.org/2000/svg"
+            width="16"
+            height="16"
+            viewBox="0 0 24 24"
+            fill="none"
+            stroke="currentColor"
+            stroke-width="2"
+            stroke-linecap="round"
+            stroke-linejoin="round"
+          >
+            <path d="M14 2H6a2 2 0 0 0-2 2v16a2 2 0 0 0 2 2h12a2 2 0 0 0 2-2V8z"></path>
+            <polyline points="14,2 14,8 20,8"></polyline>
+            <line x1="16" y1="13" x2="8" y2="13"></line>
+            <line x1="16" y1="17" x2="8" y2="17"></line>
+            <polyline points="10,9 9,9 8,9"></polyline>
+          </svg>
+        </span>
+      );
+    default:
+      return null;
+  }
+};
+
+const renderCost = (cost?: number) => {
+  return cost === undefined ? "-" : `$${cost.toFixed(2)}`;
+};
 
 export const Rendered = renderToString(
   <Fragment>
@@ -138,8 +213,7 @@ export const Rendered = renderToString(
         <button id="help">How to use</button>
       </div>
     </header>
-    <div id="table-viewport" class="table-viewport">
-      <table id="models-table">
+    <table>
       <thead>
         <tr>
           <th class="sortable" data-type="text">
@@ -268,12 +342,120 @@ export const Rendered = renderToString(
           </th>
         </tr>
       </thead>
-      <tbody id="models-table-body" dangerouslySetInnerHTML={{
-        __html: TableRows.slice(0, INITIAL_ROW_COUNT).map((row, i) => renderRow(row, i)).join('')
-          + renderRow(largestRow, -1).replace('<tr', '<tr style="visibility:hidden" aria-hidden="true"')
-      }} />
-      </table>
-    </div>
+      <tbody>
+        {Object.entries(Providers)
+          .sort(([, providerA], [, providerB]) =>
+            providerA.name.localeCompare(providerB.name)
+          )
+          .flatMap(([providerId, provider]) =>
+            Object.entries(provider.models)
+              .filter(([, model]) => model.status !== "alpha")
+              .sort(([, modelA], [, modelB]) =>
+                modelA.name.localeCompare(modelB.name)
+              )
+              .map(([modelId, model]) => (
+                <tr key={`${providerId}-${modelId}`}>
+                  <td>
+                    <div class="provider-cell">
+                      {renderProviderLogo(providerId)}
+                      <span>{provider.name}</span>
+                    </div>
+                  </td>
+                  <td>{model.name}</td>
+                  <td>{model.family ?? "-"}</td>
+                  <td>{providerId}</td>
+                  <td>
+                    <div class="model-id-cell">
+                      <span class="model-id-text">{modelId}</span>
+                      <button
+                        class="copy-button"
+                        onclick={`copyModelId(this, '${modelId}')`}
+                      >
+                        <svg
+                          class="copy-icon"
+                          xmlns="http://www.w3.org/2000/svg"
+                          width="14"
+                          height="14"
+                          viewBox="0 0 24 24"
+                          fill="none"
+                          stroke="currentColor"
+                          stroke-width="2"
+                          stroke-linecap="round"
+                          stroke-linejoin="round"
+                        >
+                          <rect
+                            width="14"
+                            height="14"
+                            x="8"
+                            y="8"
+                            rx="2"
+                            ry="2"
+                          />
+                          <path d="m4 16c-1.1 0-2-.9-2-2V4c0-1.1.9-2 2-2h10c1.1 0 2 .9 2 2" />
+                        </svg>
+                        <svg
+                          class="check-icon"
+                          xmlns="http://www.w3.org/2000/svg"
+                          width="14"
+                          height="14"
+                          viewBox="0 0 24 24"
+                          fill="none"
+                          stroke="currentColor"
+                          stroke-width="2"
+                          stroke-linecap="round"
+                          stroke-linejoin="round"
+                          style="display: none;"
+                        >
+                          <polyline points="20,6 9,17 4,12" />
+                        </svg>
+                      </button>
+                    </div>
+                  </td>
+                  <td>{model.tool_call ? "Yes" : "No"}</td>
+                  <td>{model.reasoning ? "Yes" : "No"}</td>
+                  <td>
+                    <div class="modalities">
+                      {model.modalities.input.map((modality) =>
+                        getModalityIcon(modality)
+                      )}
+                    </div>
+                  </td>
+                  <td>
+                    <div class="modalities">
+                      {model.modalities.output.map((modality) =>
+                        getModalityIcon(modality)
+                      )}
+                    </div>
+                  </td>
+                  <td>{renderCost(model.cost?.input)}</td>
+                  <td>{renderCost(model.cost?.output)}</td>
+                  <td>{renderCost(model.cost?.reasoning)}</td>
+                  <td>{renderCost(model.cost?.cache_read)}</td>
+                  <td>{renderCost(model.cost?.cache_write)}</td>
+                  <td>{renderCost(model.cost?.input_audio)}</td>
+                  <td>{renderCost(model.cost?.output_audio)}</td>
+                  <td>{model.limit.context.toLocaleString()}</td>
+                  <td>{model.limit.input?.toLocaleString() ?? "-"}</td>
+                  <td>{model.limit.output.toLocaleString()}</td>
+                  <td>
+                    {model.structured_output === undefined
+                      ? "-"
+                      : model.structured_output
+                      ? "Yes"
+                      : "No"}
+                  </td>
+                  <td>{model.temperature ? "Yes" : "No"}</td>
+                  <td>{model.open_weights ? "Open" : "Closed"}</td>
+                  <td>
+                    {model.knowledge ? model.knowledge.substring(0, 7) : "-"}
+                  </td>
+                  <td>{model.release_date}</td>
+                  <td>{model.last_updated}</td>
+                </tr>
+              ))
+          )}
+      </tbody>
+    </table>
     <dialog id="modal">
       <div class="header">
         <h2>How to use</h2>
@@ -383,15 +565,10 @@ export const Rendered = renderToString(
         >
           Edit on GitHub
         </a>
-        <a href="https://opencode.ai" target="_blank" rel="noopener noreferrer">
-          Created by OpenCode
+        <a href="https://sst.dev" target="_blank" rel="noopener noreferrer">
+          Created by SST
         </a>
       </div>
     </dialog>
-    <script
-      dangerouslySetInnerHTML={{
-        __html: `window.__TABLE_DATA__ = ${JSON.stringify(TableRows)}`,
-      }}
-    ></script>
   </Fragment>
 );

packages/web/src/shared.ts

@@ -1,176 +0,0 @@
-export interface TableRow {
-  providerId: string;
-  providerName: string;
-  providerLogoSvg: string;
-  modelId: string;
-  modelName: string;
-  family?: string;
-  toolCall: boolean;
-  reasoning: boolean;
-  input: string[];
-  output: string[];
-  inputCost?: number;
-  outputCost?: number;
-  reasoningCost?: number;
-  cacheReadCost?: number;
-  cacheWriteCost?: number;
-  audioInputCost?: number;
-  audioOutputCost?: number;
-  contextLimit: number;
-  inputLimit?: number;
-  outputLimit: number;
-  structuredOutput?: boolean;
-  temperature: boolean;
-  openWeights: boolean;
-  knowledge?: string;
-  releaseDate: string;
-  lastUpdated: string;
-}
-
-const MODALITY_ICONS: Record<string, string> = {
-  text: `<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><polyline points="4,7 4,4 20,4 20,7"></polyline><line x1="9" y1="20" x2="15" y2="20"></line><line x1="12" y1="4" x2="12" y2="20"></line></svg>`,
-  image: `<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><rect width="18" height="18" x="3" y="3" rx="2" ry="2"></rect><circle cx="9" cy="9" r="2"></circle><path d="m21 15-3.086-3.086a2 2 0 0 0-2.828 0L6 21"></path></svg>`,
-  audio: `<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><polygon points="11 5 6 9 2 9 2 15 6 15 11 19 11 5"></polygon><path d="m19.07 4.93a10 10 0 0 1 0 14.14M15.54 8.46a5 5 0 0 1 0 7.07"></path></svg>`,
-  video: `<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="m22 8-6 4 6 4V8Z"></path><rect width="14" height="12" x="2" y="6" rx="2" ry="2"></rect></svg>`,
-  pdf: `<svg xmlns="http://www.w3.org/2000/svg" width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><path d="M14 2H6a2 2 0 0 0-2 2v16a2 2 0 0 0 2 2h12a2 2 0 0 0 2-2V8z"></path><polyline points="14,2 14,8 20,8"></polyline><line x1="16" y1="13" x2="8" y2="13"></line><line x1="16" y1="17" x2="8" y2="17"></line><polyline points="10,9 9,9 8,9"></polyline></svg>`,
-};
-
-export function escapeHtml(value: string | number) {
-  return String(value).replace(/[&<>'"]/g, (char) => {
-    switch (char) {
-      case "&":
-        return "&amp;";
-      case "<":
-        return "&lt;";
-      case ">":
-        return "&gt;";
-      case "'":
-        return "&#39;";
-      case '"':
-        return "&quot;";
-      default:
-        return char;
-    }
-  });
-}
-
-export function booleanText(value: boolean) {
-  return value ? "Yes" : "No";
-}
-
-export function optionalBooleanText(value?: boolean) {
-  return value === undefined ? "-" : booleanText(value);
-}
-
-export function formatCost(cost?: number) {
-  return cost === undefined ? "-" : `$${cost.toFixed(2)}`;
-}
-
-export function formatNumber(value?: number) {
-  return value === undefined ? "-" : value.toLocaleString();
-}
-
-export function knowledgeText(value?: string) {
-  return value ? value.substring(0, 7) : "-";
-}
-
-export function weightsText(value: boolean) {
-  return value ? "Open" : "Closed";
-}
-
-export function renderModalityIcon(modality: string) {
-  const label =
-    modality === "pdf"
-      ? "PDF"
-      : modality[0]!.toUpperCase() + modality.slice(1);
-  const icon = MODALITY_ICONS[modality];
-  if (!icon) return "";
-  return `<span class="modality-icon" data-tooltip="${label}">${icon}</span>`;
-}
-
-export function renderModalities(modalities: string[]) {
-  return `<div class="modalities">${modalities
-    .map(renderModalityIcon)
-    .join("")}</div>`;
-}
-
-export function renderCopyButton(modelId: string) {
-  const escapedModelId = escapeHtml(modelId);
-  return `<button type="button" class="copy-button" data-model-id="${escapedModelId}" aria-label="Copy model ID"><svg class="copy-icon" xmlns="http://www.w3.org/2000/svg" width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round"><rect width="14" height="14" x="8" y="8" rx="2" ry="2"></rect><path d="m4 16c-1.1 0-2-.9-2-2V4c0-1.1.9-2 2-2h10c1.1 0 2 .9 2 2"></path></svg><svg class="check-icon" xmlns="http://www.w3.org/2000/svg" width="14" height="14" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round" style="display: none;"><polyline points="20,6 9,17 4,12"></polyline></svg></button>`;
-}
-
-export function renderRow(row: TableRow, index: number) {
-  return `<tr data-index="${index}">
-    <td><div class="provider-cell">${row.providerLogoSvg}<span>${escapeHtml(
-    row.providerName
-  )}</span></div></td>
-    <td>${escapeHtml(row.modelName)}</td>
-    <td>${escapeHtml(row.family ?? "-")}</td>
-    <td>${escapeHtml(row.providerId)}</td>
-    <td><div class="model-id-cell"><span class="model-id-text">${escapeHtml(
-      row.modelId
-    )}</span>${renderCopyButton(row.modelId)}</div></td>
-    <td>${booleanText(row.toolCall)}</td>
-    <td>${booleanText(row.reasoning)}</td>
-    <td>${renderModalities(row.input)}</td>
-    <td>${renderModalities(row.output)}</td>
-    <td>${formatCost(row.inputCost)}</td>
-    <td>${formatCost(row.outputCost)}</td>
-    <td>${formatCost(row.reasoningCost)}</td>
-    <td>${formatCost(row.cacheReadCost)}</td>
-    <td>${formatCost(row.cacheWriteCost)}</td>
-    <td>${formatCost(row.audioInputCost)}</td>
-    <td>${formatCost(row.audioOutputCost)}</td>
-    <td>${formatNumber(row.contextLimit)}</td>
-    <td>${formatNumber(row.inputLimit)}</td>
-    <td>${formatNumber(row.outputLimit)}</td>
-    <td>${optionalBooleanText(row.structuredOutput)}</td>
-    <td>${booleanText(row.temperature)}</td>
-    <td>${weightsText(row.openWeights)}</td>
-    <td>${knowledgeText(row.knowledge)}</td>
-    <td>${escapeHtml(row.releaseDate)}</td>
-    <td>${escapeHtml(row.lastUpdated)}</td>
-  </tr>`;
-}
-
-export function getLargestRow(rows: TableRow[]): TableRow {
-  const worst: TableRow = {
-    providerId: "", providerName: "", providerLogoSvg: "", modelId: "", modelName: "",
-    toolCall: true, reasoning: true,
-    input: [], output: [],
-    contextLimit: 0, outputLimit: 0,
-    structuredOutput: true, temperature: true, openWeights: false,
-    releaseDate: "", lastUpdated: "",
-  };
-
-  for (const row of rows) {
-    if (row.providerName.length > worst.providerName.length) worst.providerName = row.providerName;
-    if (row.modelName.length > worst.modelName.length) worst.modelName = row.modelName;
-    if ((row.family ?? "").length > (worst.family ?? "").length) worst.family = row.family;
-    if (row.providerId.length > worst.providerId.length) worst.providerId = row.providerId;
-    if (row.modelId.length > worst.modelId.length) worst.modelId = row.modelId;
-    if ((row.knowledge ?? "").length > (worst.knowledge ?? "").length) worst.knowledge = row.knowledge;
-    if (row.releaseDate.length > worst.releaseDate.length) worst.releaseDate = row.releaseDate;
-    if (row.lastUpdated.length > worst.lastUpdated.length) worst.lastUpdated = row.lastUpdated;
-    if (row.input.length > worst.input.length) worst.input = row.input;
-    if (row.output.length > worst.output.length) worst.output = row.output;
-
-    const costWider = (a: number | undefined, b: number | undefined) =>
-      b !== undefined && (a === undefined || formatCost(b).length > formatCost(a).length);
-    if (costWider(worst.inputCost, row.inputCost)) worst.inputCost = row.inputCost;
-    if (costWider(worst.outputCost, row.outputCost)) worst.outputCost = row.outputCost;
-    if (costWider(worst.reasoningCost, row.reasoningCost)) worst.reasoningCost = row.reasoningCost;
-    if (costWider(worst.cacheReadCost, row.cacheReadCost)) worst.cacheReadCost = row.cacheReadCost;
-    if (costWider(worst.cacheWriteCost, row.cacheWriteCost)) worst.cacheWriteCost = row.cacheWriteCost;
-    if (costWider(worst.audioInputCost, row.audioInputCost)) worst.audioInputCost = row.audioInputCost;
-    if (costWider(worst.audioOutputCost, row.audioOutputCost)) worst.audioOutputCost = row.audioOutputCost;
-
-    const numWider = (a: number | undefined, b: number | undefined) =>
-      b !== undefined && (a === undefined || formatNumber(b).length > formatNumber(a).length);
-    if (numWider(worst.contextLimit as number | undefined, row.contextLimit)) worst.contextLimit = row.contextLimit;
-    if (numWider(worst.inputLimit, row.inputLimit)) worst.inputLimit = row.inputLimit;
-    if (numWider(worst.outputLimit as number | undefined, row.outputLimit)) worst.outputLimit = row.outputLimit;
-  }
-
-  return worst;
-}

providers/302ai/models/claude-opus-4-7.toml

@@ -15,7 +15,8 @@ output = 25.000
 cache_read = 0.500
 cache_write = 6.250
 
-[cost.context_over_200k]
+[[cost.tiers]]
+tier = { size = 200_000 }
 input = 10.000
 output = 37.500
 cache_read = 1.000

providers/302ai/models/gpt-5.4-pro.toml

@@ -16,7 +16,8 @@ output = 180.000
 cache_read = 0
 cache_write = 0
 
-[cost.context_over_200k]
+[[cost.tiers]]
+tier = { size = 272_000 }
 input = 60.000
 output = 270.000
 

providers/302ai/models/gpt-5.4.toml

@@ -16,7 +16,8 @@ output = 15.000
 cache_read = 0.250
 cache_write = 0
 
-[cost.context_over_200k]
+[[cost.tiers]]
+tier = { size = 272_000 }
 input = 5.000
 output = 22.500
 

providers/aihubmix/models/claude-opus-4-7-think.toml

@@ -0,0 +1,25 @@
+name = "Claude Opus 4.7 Thinking"
+family = "claude-opus"
+release_date = "2026-04-16"
+last_updated = "2026-04-16"
+attachment = true
+reasoning = true
+temperature = false
+tool_call = true
+structured_output = true
+knowledge = "2026-01-31"
+open_weights = false
+
+[cost]
+input = 5
+output = 25
+cache_read = 0.5
+cache_write = 6.25
+
+[limit]
+context = 200_000
+output = 32_000
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

providers/aihubmix/models/claude-sonnet-4-6-think.toml

@@ -15,7 +15,8 @@ output = 15.00
 cache_read = 0.30
 cache_write = 3.75
 
-[cost.context_over_200k]
+[[cost.tiers]]
+tier = { size = 200_000 }
 input = 6.00
 output = 22.50
 cache_read = 0.60

providers/aihubmix/models/claude-sonnet-4-6.toml

@@ -15,7 +15,8 @@ output = 15.00
 cache_read = 0.30
 cache_write = 3.75
 
-[cost.context_over_200k]
+[[cost.tiers]]
+tier = { size = 200_000 }
 input = 6.00
 output = 22.50
 cache_read = 0.60

providers/aihubmix/models/coding-glm-5-free.toml

@@ -1,24 +0,0 @@
-name = "Coding-GLM-5-Free"
-family = "glm"
-release_date = "2026-02-11"
-last_updated = "2026-02-11"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-open_weights = false
-
-[interleaved]
-field = "reasoning_content"
-
-[cost]
-input = 0.0
-output = 0.0
-
-[limit]
-context = 204800
-output = 131072
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/aihubmix/models/coding-glm-5.1-free.toml

@@ -1,25 +1,26 @@
-name = "GLM 4.5 Air"
+name = "Coding GLM 5.1 (free)"
 family = "glm"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
+release_date = "2026-04-11"
+last_updated = "2026-04-11"
 attachment = false
 reasoning = true
 temperature = true
 tool_call = true
 structured_output = true
-open_weights = true
+open_weights = false
+
+[interleaved]
+field = "reasoning_content"
 
 [cost]
-input = 0.05
-output = 0.22
+input = 0
+output = 0
+cache_read = 0
 
 [limit]
-context = 131_072
-output = 131_072
+context = 204_800
+output = 128_000
 
 [modalities]
 input = ["text"]
 output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/aihubmix/models/coding-minimax-m2.7-highspeed.toml

@@ -1,23 +1,21 @@
-name = "MiniMax-M2.1"
+name = "Coding MiniMax M2.7 Highspeed"
 family = "minimax"
-release_date = "2025-12-23"
-last_updated = "2025-12-23"
+release_date = "2026-03-18"
+last_updated = "2026-03-18"
 attachment = false
 reasoning = true
 temperature = true
 tool_call = true
+structured_output = true
 open_weights = true
 
-[interleaved]
-field = "reasoning_details"
-
 [cost]
-input = 0.288
-output = 1.152
+input = 0.2
+output = 0.2
 
 [limit]
 context = 204_800
-output = 192_000
+output = 13_100
 
 [modalities]
 input = ["text"]

providers/aihubmix/models/coding-minimax-m2.7.toml

@@ -1,20 +1,21 @@
-name = "MiniMax-M2.1"
+name = "Coding MiniMax M2.7"
 family = "minimax"
-release_date = "2025-12-23"
-last_updated = "2025-12-23"
+release_date = "2026-03-18"
+last_updated = "2026-03-18"
 attachment = false
 reasoning = true
 temperature = true
 tool_call = true
+structured_output = true
 open_weights = true
 
 [cost]
-input = 0.0
-output = 0.0
+input = 0.2
+output = 0.2
 
 [limit]
 context = 204_800
-output = 131_072
+output = 13_100
 
 [modalities]
 input = ["text"]

providers/aihubmix/models/gemini-3.1-flash-lite.toml

@@ -1,4 +1,4 @@
-name = "Gemini 3.1 Flash Lite Preview"
+name = "Gemini 3.1 Flash Lite"
 family = "gemini-flash-lite"
 release_date = "2026-03-03"
 last_updated = "2026-03-03"

providers/aihubmix/models/gemini-3.1-pro-preview-customtools.toml

@@ -0,0 +1,30 @@
+name = "Gemini 3.1 Pro Preview Custom Tools"
+family = "gemini-pro"
+release_date = "2026-02-19"
+last_updated = "2026-02-19"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[cost]
+input = 2
+output = 12
+cache_read = 0.2
+
+[[cost.tiers]]
+tier = { size = 200_000 }
+input = 4
+output = 18
+cache_read = 0.4
+
+[limit]
+context = 1_048_576
+output = 65_536
+
+[modalities]
+input = ["text", "image", "audio", "video", "pdf"]
+output = ["text"]

providers/aihubmix/models/glm-5.toml

@@ -1,25 +0,0 @@
-name = "GLM-5"
-family = "glm"
-release_date = "2026-02-11"
-last_updated = "2026-02-11"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-open_weights = true
-
-[interleaved]
-field = "reasoning_content"
-
-[cost]
-input = 0.88
-output = 2.816
-cache_read = 0.176
-
-[limit]
-context = 202_752
-output = 0
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/aihubmix/models/glm-5v-turbo.toml

@@ -0,0 +1,26 @@
+name = "GLM 5 Vision Turbo"
+family = "glm"
+release_date = "2026-05-09"
+last_updated = "2026-05-09"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = false
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 0.7042
+output = 3.09848
+cache_read = 0.169008
+
+[limit]
+context = 200_000
+output = 128_000
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]

providers/aihubmix/models/grok-4.3.toml

@@ -0,0 +1,29 @@
+name = "Grok 4.3"
+family = "grok"
+release_date = "2026-05-01"
+last_updated = "2026-05-01"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = false
+
+[cost]
+input = 1.25
+output = 2.5
+cache_read = 0.2
+
+[[cost.tiers]]
+tier = { size = 200_000 }
+input = 2.5
+output = 5.0
+cache_read = 0.4
+
+[limit]
+context = 1_000_000
+output = 1_000_000
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

providers/aihubmix/models/qwen3-coder-next.toml

@@ -1,24 +0,0 @@
-name = "Qwen3 Coder Plus"
-family = "qwen"
-release_date = "2025-07-23"
-last_updated = "2025-07-23"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-knowledge = "2025-04"
-open_weights = true
-
-[cost]
-input = 0.137
-output = 0.548
-cache_read = 0.137
-
-[limit]
-context = 2_000_000
-output = 64_000
-input = 262_144
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/aihubmix/models/qwen3-max-2026-01-23.toml

@@ -1,23 +0,0 @@
-name = "Qwen3 Max"
-family = "qwen"
-release_date = "2025-09-23"
-last_updated = "2025-09-23"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-knowledge = "2025-04"
-open_weights = false
-
-[cost]
-input = 0.34246
-output = 1.36984
-cache_read = 0.34246
-
-[limit]
-context = 252_000
-output = 32_000
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/aihubmix/models/qwen3.6-flash.toml

@@ -1,11 +1,12 @@
-name = "Qwen3.6 Plus"
+name = "Qwen3.6 Flash"
 family = "qwen"
 release_date = "2026-04-02"
 last_updated = "2026-04-02"
-attachment = false
+attachment = true
 reasoning = true
 temperature = true
 tool_call = true
+structured_output = true
 knowledge = "2025-04"
 open_weights = false
 
@@ -16,8 +17,8 @@ cache_read = 0.0169
 cache_write = 0.21125
 
 [limit]
-context = 1_000_000
-output = 65_536
+context = 991_000
+output = 64_000
 
 [modalities]
 input = ["text", "image", "video"]

providers/aihubmix/models/qwen3.6-max-preview.toml

@@ -0,0 +1,25 @@
+name = "Qwen3.6 Max Preview"
+family = "qwen"
+release_date = "2026-05-09"
+last_updated = "2026-05-09"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-04"
+open_weights = false
+
+[cost]
+input = 1.268
+output = 7.608
+cache_read = 0.1268
+cache_write = 1.585
+
+[limit]
+context = 240_000
+output = 64_000
+
+[modalities]
+input = ["text"]
+output = ["text"]

providers/aihubmix/models/qwen3.6-plus.toml

@@ -1,19 +1,20 @@
-name = "Qwen3.5 Plus"
+name = "Qwen3.6 Plus"
 family = "qwen"
-release_date = "2026-02-16"
-last_updated = "2026-02-16"
-attachment = false
+release_date = "2026-05-09"
+last_updated = "2026-05-09"
+attachment = true
 reasoning = true
 temperature = true
 tool_call = true
+structured_output = true
 knowledge = "2025-04"
 open_weights = false
 
 [cost]
-input = 0.1096
-output = 0.6576
-cache_read = 0.01096
-cache_write = 0.137
+input = 0.282
+output = 1.692
+cache_read = 0.0282
+cache_write = 0.3525
 
 [limit]
 context = 991_000

providers/alibaba-cn/models/qwen3.6-plus.toml

@@ -10,10 +10,17 @@ tool_call = true
 open_weights = false
 
 [cost]
-input = 0.276
-output = 1.651
-cache_read = 0.028
-cache_write = 0.344
+input = 0.50
+output = 3.00
+cache_read = 0.05
+cache_write = 0.625
+
+[[cost.tiers]]
+tier = { size = 256_000 }
+input = 2.00
+output = 6.00
+cache_read = 0.20
+cache_write = 2.50
 
 [limit]
 context = 1_000_000

providers/alibaba/models/qwen3.6-plus.toml

@@ -10,10 +10,17 @@ tool_call = true
 open_weights = false
 
 [cost]
-input = 0.276
-output = 1.651
-cache_read = 0.028
-cache_write = 0.344
+input = 0.50
+output = 3.00
+cache_read = 0.05
+cache_write = 0.625
+
+[[cost.tiers]]
+tier = { size = 256_000 }
+input = 2.00
+output = 6.00
+cache_read = 0.20
+cache_write = 2.50
 
 [limit]
 context = 1_000_000

providers/amazon-bedrock/models/amazon.nova-premier-v1:0.toml

@@ -1,22 +0,0 @@
-name = "Nova Premier"
-family = "nova"
-release_date = "2024-12-03"
-last_updated = "2024-12-03"
-attachment = true
-reasoning = true
-temperature = true
-knowledge = "2024-10"
-tool_call = true
-open_weights = false
-
-[cost]
-input = 2.50
-output = 12.50
-
-[limit]
-context = 1_000_000
-output = 16_384
-
-[modalities]
-input = ["text", "image", "video"]
-output = ["text"]

providers/amazon-bedrock/models/anthropic.claude-3-5-haiku-20241022-v1:0.toml

@@ -1,2 +0,0 @@
-[extends]
-from = "anthropic/claude-3-5-haiku-20241022"

providers/amazon-bedrock/models/anthropic.claude-3-5-sonnet-20240620-v1:0.toml

@@ -1,2 +0,0 @@
-[extends]
-from = "anthropic/claude-3-5-sonnet-20240620"

providers/amazon-bedrock/models/anthropic.claude-3-5-sonnet-20241022-v2:0.toml

@@ -1,24 +0,0 @@
-name = "Claude Sonnet 3.5 v2"
-family = "claude-sonnet"
-release_date = "2024-10-22"
-last_updated = "2024-10-22"
-attachment = true
-reasoning = false
-temperature = true
-knowledge = "2024-04"
-tool_call = true
-open_weights = false
-
-[cost]
-input = 3.00
-output = 15.00
-cache_read = 0.30
-cache_write = 3.75
-
-[limit]
-context = 200_000
-output = 8_192
-
-[modalities]
-input = ["text", "image", "pdf"]
-output = ["text"]

providers/amazon-bedrock/models/anthropic.claude-3-7-sonnet-20250219-v1:0.toml

@@ -1,24 +0,0 @@
-name = "Claude Sonnet 3.7"
-family = "claude-sonnet"
-release_date = "2025-02-19"
-last_updated = "2025-02-19"
-attachment = true
-reasoning = false
-temperature = true
-knowledge = "2024-04"
-tool_call = true
-open_weights = false
-
-[cost]
-input = 3.00
-output = 15.00
-cache_read = 0.30
-cache_write = 3.75
-
-[limit]
-context = 200_000
-output = 8_192
-
-[modalities]
-input = ["text", "image", "pdf"]
-output = ["text"]

providers/amazon-bedrock/models/anthropic.claude-3-haiku-20240307-v1:0.toml

@@ -1,22 +0,0 @@
-name = "Claude Haiku 3"
-family = "claude-haiku"
-release_date = "2024-03-13"
-last_updated = "2024-03-13"
-attachment = true
-reasoning = false
-temperature = true
-knowledge = "2024-02"
-tool_call = true
-open_weights = false
-
-[cost]
-input = 0.25
-output = 1.25
-
-[limit]
-context = 200_000
-output = 4_096
-
-[modalities]
-input = ["text", "image", "pdf"]
-output = ["text"]

providers/amazon-bedrock/models/anthropic.claude-opus-4-20250514-v1:0.toml

@@ -1,2 +0,0 @@
-[extends]
-from = "anthropic/claude-opus-4-20250514"

providers/amazon-bedrock/models/anthropic.claude-sonnet-4-20250514-v1:0.toml

@@ -1,2 +0,0 @@
-[extends]
-from = "anthropic/claude-sonnet-4-20250514"

providers/amazon-bedrock/models/au.anthropic.claude-haiku-4-5-20251001-v1:0.toml

@@ -0,0 +1,4 @@
+name = "Claude Haiku 4.5 (AU)"
+
+[extends]
+from = "anthropic/claude-haiku-4-5-20251001"

providers/amazon-bedrock/models/au.anthropic.claude-sonnet-4-5-20250929-v1:0.toml

@@ -0,0 +1,4 @@
+name = "Claude Sonnet 4.5 (AU)"
+
+[extends]
+from = "anthropic/claude-sonnet-4-5"

providers/amazon-bedrock/models/eu.anthropic.claude-sonnet-4-20250514-v1:0.toml

@@ -1,4 +0,0 @@
-name = "Claude Sonnet 4 (EU)"
-
-[extends]
-from = "anthropic/claude-sonnet-4-20250514"

providers/amazon-bedrock/models/global.anthropic.claude-sonnet-4-20250514-v1:0.toml

@@ -1,4 +0,0 @@
-name = "Claude Sonnet 4 (Global)"
-
-[extends]
-from = "anthropic/claude-sonnet-4-20250514"

providers/amazon-bedrock/models/jp.anthropic.claude-opus-4-7.toml

@@ -0,0 +1,4 @@
+name = "Claude Opus 4.7 (JP)"
+
+[extends]
+from = "anthropic/claude-opus-4-7"

providers/amazon-bedrock/models/jp.anthropic.claude-sonnet-4-5-20250929-v1:0.toml

@@ -0,0 +1,4 @@
+name = "Claude Sonnet 4.5 (JP)"
+
+[extends]
+from = "anthropic/claude-sonnet-4-5"

providers/amazon-bedrock/models/jp.anthropic.claude-sonnet-4-6.toml

@@ -0,0 +1,4 @@
+name = "Claude Sonnet 4.6 (JP)"
+
+[extends]
+from = "anthropic/claude-sonnet-4-6"

providers/amazon-bedrock/models/us.anthropic.claude-opus-4-20250514-v1:0.toml

@@ -1,4 +0,0 @@
-name = "Claude Opus 4 (US)"
-
-[extends]
-from = "anthropic/claude-opus-4-20250514"

providers/amazon-bedrock/models/us.anthropic.claude-sonnet-4-20250514-v1:0.toml

@@ -1,4 +0,0 @@
-name = "Claude Sonnet 4 (US)"
-
-[extends]
-from = "anthropic/claude-sonnet-4-20250514"

providers/amazon-bedrock/models/us.deepseek.r1-v1:0.toml

@@ -0,0 +1,4 @@
+name = "DeepSeek-R1 (US)"
+
+[extends]
+from = "amazon-bedrock/deepseek.r1-v1:0"

providers/amazon-bedrock/models/us.meta.llama4-maverick-17b-instruct-v1:0.toml

@@ -0,0 +1,4 @@
+name = "Llama 4 Maverick 17B Instruct (US)"
+
+[extends]
+from = "amazon-bedrock/meta.llama4-maverick-17b-instruct-v1:0"

providers/amazon-bedrock/models/us.meta.llama4-scout-17b-instruct-v1:0.toml

@@ -0,0 +1,4 @@
+name = "Llama 4 Scout 17B Instruct (US)"
+
+[extends]
+from = "amazon-bedrock/meta.llama4-scout-17b-instruct-v1:0"

providers/azure-cognitive-services/models/claude-opus-4-6.toml

@@ -15,7 +15,8 @@ output = 25.00
 cache_read = 0.50
 cache_write = 6.25
 
-[cost.context_over_200k]
+[[cost.tiers]]
+tier = { size = 200_000 }
 input = 10.00
 output = 37.50
 cache_read = 1.00

providers/azure/models/claude-opus-4-6.toml

@@ -15,7 +15,8 @@ output = 25.00
 cache_read = 0.50
 cache_write = 6.25
 
-[cost.context_over_200k]
+[[cost.tiers]]
+tier = { size = 200_000 }
 input = 10.00
 output = 37.50
 cache_read = 1.00

providers/chutes/models/MiniMaxAI/MiniMax-M2.1-TEE.toml

@@ -1,25 +0,0 @@
-name = "MiniMax M2.1 TEE"
-family = "minimax"
-release_date = "2025-12-29"
-last_updated = "2026-01-27"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.27
-output = 1.12
-
-[limit]
-context = 196_608
-output = 65_536
-
-[modalities]
-input = ["text"]
-output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/chutes/models/MiniMaxAI/MiniMax-M2.5-TEE.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "MiniMax M2.5 TEE"
 family = "minimax"
 release_date = "2026-02-15"
-last_updated = "2026-02-15"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true
@@ -10,9 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.30
-output = 1.10
-cache_read = 0.15
+input = 0.15
+output = 1.2
+cache_read = 0.075
 
 [limit]
 context = 196_608

providers/chutes/models/NousResearch/DeepHermes-3-Mistral-24B-Preview.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "DeepHermes 3 Mistral 24B Preview"
 family = "nousresearch"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = false
 temperature = true
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.02
-output = 0.10
+input = 0.0245
+output = 0.0978
+cache_read = 0.01225
 
 [limit]
 context = 32_768

providers/chutes/models/NousResearch/Hermes-4-14B.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Hermes 4 14B"
 family = "nousresearch"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.01
-output = 0.05
+input = 0.0136
+output = 0.0543
+cache_read = 0.0068
 
 [limit]
 context = 40_960

providers/chutes/models/NousResearch/Hermes-4-405B-FP8-TEE.toml

@@ -1,25 +0,0 @@
-name = "Hermes 4 405B FP8 TEE"
-family = "nousresearch"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.30
-output = 1.20
-
-[limit]
-context = 131_072
-output = 65_536
-
-[modalities]
-input = ["text"]
-output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/chutes/models/NousResearch/Hermes-4-70B.toml

@@ -1,25 +0,0 @@
-name = "Hermes 4 70B"
-family = "nousresearch"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.11
-output = 0.38
-
-[limit]
-context = 131_072
-output = 131_072
-
-[modalities]
-input = ["text"]
-output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/chutes/models/OpenGVLab/InternVL3-78B-TEE.toml

@@ -1,22 +0,0 @@
-name = "InternVL3 78B TEE"
-family = "opengvlab"
-release_date = "2025-01-06"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = false
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.10
-output = 0.39
-
-[limit]
-context = 32_768
-output = 32_768
-
-[modalities]
-input = ["text", "image"]
-output = ["text"]

providers/chutes/models/Qwen/Qwen2.5-72B-Instruct.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen2.5 72B Instruct"
 family = "qwen"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = false
 temperature = true
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.13
-output = 0.52
+input = 0.2989
+output = 1.1957
+cache_read = 0.14945
 
 [limit]
 context = 32_768

providers/chutes/models/Qwen/Qwen2.5-Coder-32B-Instruct.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen2.5 Coder 32B Instruct"
 family = "qwen"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = false
 temperature = true
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.03
-output = 0.11
+input = 0.0272
+output = 0.1087
+cache_read = 0.0136
 
 [limit]
 context = 32_768

providers/chutes/models/Qwen/Qwen2.5-VL-32B-Instruct.toml

@@ -1,8 +1,10 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen2.5 VL 32B Instruct"
 family = "qwen"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
+last_updated = "2026-04-25"
+attachment = true
 reasoning = false
 temperature = true
 tool_call = false
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.05
-output = 0.22
+input = 0.0543
+output = 0.2174
+cache_read = 0.02715
 
 [limit]
 context = 16_384

providers/chutes/models/Qwen/Qwen2.5-VL-72B-Instruct-TEE.toml

@@ -1,22 +0,0 @@
-name = "Qwen2.5 VL 72B Instruct TEE"
-family = "qwen"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = false
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.15
-output = 0.60
-
-[limit]
-context = 32_768
-output = 32_768
-
-[modalities]
-input = ["text", "image"]
-output = ["text"]

providers/chutes/models/Qwen/Qwen3-235B-A22B-Instruct-2507-TEE.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen3 235B A22B Instruct 2507 TEE"
 family = "qwen"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = false
 temperature = true
@@ -10,9 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.08
-output = 0.55
-cache_read = 0.04
+input = 0.1
+output = 0.6
+cache_read = 0.05
 
 [limit]
 context = 262_144

providers/chutes/models/Qwen/Qwen3-235B-A22B-Thinking-2507.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen3 235B A22B Thinking 2507"
 family = "qwen"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true
@@ -11,7 +13,8 @@ open_weights = true
 
 [cost]
 input = 0.11
-output = 0.60
+output = 0.6
+cache_read = 0.055
 
 [limit]
 context = 262_144

providers/chutes/models/Qwen/Qwen3-235B-A22B.toml

@@ -1,25 +0,0 @@
-name = "Qwen3 235B A22B"
-family = "qwen"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.30
-output = 1.20
-
-[limit]
-context = 40_960
-output = 40_960
-
-[modalities]
-input = ["text"]
-output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/chutes/models/Qwen/Qwen3-30B-A3B-Instruct-2507.toml

@@ -1,22 +0,0 @@
-name = "Qwen3 30B A3B Instruct 2507"
-family = "qwen"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.08
-output = 0.33
-
-[limit]
-context = 262_144
-output = 262_144
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/chutes/models/Qwen/Qwen3-30B-A3B.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen3 30B A3B"
 family = "qwen"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true
@@ -12,6 +14,7 @@ open_weights = true
 [cost]
 input = 0.06
 output = 0.22
+cache_read = 0.03
 
 [limit]
 context = 40_960

providers/chutes/models/Qwen/Qwen3-32B-TEE.toml

@@ -1,7 +1,9 @@
-name = "Qwen3 32B"
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
+name = "Qwen3 32B TEE"
 family = "qwen"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
+release_date = "2026-04-25"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true
@@ -21,6 +23,3 @@ output = 40_960
 [modalities]
 input = ["text"]
 output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/chutes/models/Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8-TEE.toml

@@ -1,23 +0,0 @@
-name = "Qwen3 Coder 480B A35B Instruct FP8 TEE"
-family = "qwen"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.22
-output = 0.95
-cache_read = 0.11
-
-[limit]
-context = 262_144
-output = 262_144
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/chutes/models/Qwen/Qwen3-Coder-Next-TEE.toml

@@ -0,0 +1,25 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
+name = "Qwen3 Coder Next TEE"
+family = "qwen"
+release_date = "2026-04-25"
+last_updated = "2026-04-25"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[cost]
+input = 0.12
+output = 0.75
+cache_read = 0.06
+
+[limit]
+context = 262_144
+output = 65_536
+
+[modalities]
+input = ["text"]
+output = ["text"]

providers/chutes/models/Qwen/Qwen3-Coder-Next.toml

@@ -1,22 +0,0 @@
-name = "Qwen3 Coder Next"
-family = "qwen"
-release_date = "2026-02-05"
-last_updated = "2026-02-05"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.07
-output = 0.30
-
-[limit]
-context = 262_144
-output = 65_536
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/chutes/models/Qwen/Qwen3-Next-80B-A3B-Instruct.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen3 Next 80B A3B Instruct"
 family = "qwen"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = false
 temperature = true
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.10
-output = 0.80
+input = 0.1
+output = 0.8
+cache_read = 0.05
 
 [limit]
 context = 262_144

providers/chutes/models/Qwen/Qwen3-VL-235B-A22B-Instruct.toml

@@ -1,22 +0,0 @@
-name = "Qwen3 VL 235B A22B Instruct"
-family = "qwen"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.30
-output = 1.20
-
-[limit]
-context = 262_144
-output = 262_144
-
-[modalities]
-input = ["text", "image"]
-output = ["text"]

providers/chutes/models/Qwen/Qwen3.5-397B-A17B-TEE.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen3.5 397B A17B TEE"
 family = "qwen"
 release_date = "2026-02-18"
-last_updated = "2026-02-18"
+last_updated = "2026-04-25"
 attachment = true
 reasoning = true
 temperature = true

providers/chutes/models/Qwen/Qwen3.6-27B-TEE.toml

@@ -0,0 +1,25 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
+name = "Qwen3.6 27B TEE"
+family = "qwen"
+release_date = "2026-04-25"
+last_updated = "2026-04-25"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[cost]
+input = 0.195
+output = 1.56
+cache_read = 0.0975
+
+[limit]
+context = 262_144
+output = 65_536
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

providers/chutes/models/Qwen/Qwen3Guard-Gen-0.6B.toml

@@ -1,17 +1,18 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Qwen3Guard Gen 0.6B"
 family = "qwen"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = false
 temperature = true
 tool_call = false
-structured_output = false
 open_weights = true
 
 [cost]
 input = 0.01
-output = 0.01
+output = 0.0109
 cache_read = 0.005
 
 [limit]

providers/chutes/models/XiaomiMiMo/MiMo-V2-Flash-TEE.toml

@@ -0,0 +1,25 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
+name = "MiMo V2 Flash TEE"
+family = "mimo"
+release_date = "2026-04-25"
+last_updated = "2026-04-25"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[cost]
+input = 0.09
+output = 0.29
+cache_read = 0.045
+
+[limit]
+context = 262_144
+output = 65_536
+
+[modalities]
+input = ["text"]
+output = ["text"]

providers/chutes/models/XiaomiMiMo/MiMo-V2-Flash.toml

@@ -1,22 +0,0 @@
-name = "MiMo V2 Flash"
-family = "mimo"
-release_date = "2025-12-29"
-last_updated = "2026-01-27"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = false
-open_weights = true
-
-[cost]
-input = 0.09
-output = 0.29
-
-[limit]
-context = 262_144
-output = 32_000
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/chutes/models/chutesai/Mistral-Small-3.1-24B-Instruct-2503.toml

@@ -1,23 +0,0 @@
-name = "Mistral Small 3.1 24B Instruct 2503"
-family = "chutesai"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.03
-output = 0.11
-cache_read = 0.015
-
-[limit]
-context = 131_072
-output = 131_072
-
-[modalities]
-input = ["text", "image"]
-output = ["text"]

providers/chutes/models/chutesai/Mistral-Small-3.2-24B-Instruct-2506.toml

@@ -1,22 +0,0 @@
-name = "Mistral Small 3.2 24B Instruct 2506"
-family = "chutesai"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.06
-output = 0.18
-
-[limit]
-context = 131_072
-output = 131_072
-
-[modalities]
-input = ["text", "image"]
-output = ["text"]

providers/chutes/models/deepseek-ai/DeepSeek-R1-0528-TEE.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "DeepSeek R1 0528 TEE"
 family = "deepseek-thinking"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.40
-output = 1.75
+input = 0.45
+output = 2.15
+cache_read = 0.225
 
 [limit]
 context = 163_840

providers/chutes/models/deepseek-ai/DeepSeek-R1-Distill-Llama-70B.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "DeepSeek R1 Distill Llama 70B"
 family = "deepseek-thinking"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.03
-output = 0.11
+input = 0.0272
+output = 0.1087
+cache_read = 0.0136
 
 [limit]
 context = 131_072

providers/chutes/models/deepseek-ai/DeepSeek-R1-TEE.toml

@@ -1,25 +0,0 @@
-name = "DeepSeek R1 TEE"
-family = "deepseek-thinking"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = false
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.30
-output = 1.20
-
-[limit]
-context = 163_840
-output = 163_840
-
-[modalities]
-input = ["text"]
-output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/chutes/models/deepseek-ai/DeepSeek-V3-0324-TEE.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "DeepSeek V3 0324 TEE"
 family = "deepseek"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = false
 temperature = true
@@ -10,9 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.19
-output = 0.87
-cache_read = 0.095
+input = 0.25
+output = 1
+cache_read = 0.125
 
 [limit]
 context = 163_840

providers/chutes/models/deepseek-ai/DeepSeek-V3.1-TEE.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "DeepSeek V3.1 TEE"
 family = "deepseek"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true
@@ -10,8 +12,9 @@ structured_output = true
 open_weights = true
 
 [cost]
-input = 0.20
-output = 0.80
+input = 0.27
+output = 1
+cache_read = 0.135
 
 [limit]
 context = 163_840

providers/chutes/models/deepseek-ai/DeepSeek-V3.1-Terminus-TEE.toml

@@ -1,25 +0,0 @@
-name = "DeepSeek V3.1 Terminus TEE"
-family = "deepseek"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.23
-output = 0.90
-
-[limit]
-context = 163_840
-output = 65_536
-
-[modalities]
-input = ["text"]
-output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/chutes/models/deepseek-ai/DeepSeek-V3.2-Speciale-TEE.toml

@@ -1,25 +0,0 @@
-name = "DeepSeek V3.2 Speciale TEE"
-family = "deepseek"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = false
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.27
-output = 0.41
-
-[limit]
-context = 163_840
-output = 65_536
-
-[modalities]
-input = ["text"]
-output = ["text"]
-
-[interleaved]
-field = "reasoning_content"

providers/chutes/models/deepseek-ai/DeepSeek-V3.2-TEE.toml

@@ -1,7 +1,9 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "DeepSeek V3.2 TEE"
 family = "deepseek"
 release_date = "2025-12-29"
-last_updated = "2026-01-10"
+last_updated = "2026-04-25"
 attachment = false
 reasoning = true
 temperature = true

providers/chutes/models/deepseek-ai/DeepSeek-V3.toml

@@ -1,22 +0,0 @@
-name = "DeepSeek V3"
-family = "deepseek"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = false
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.30
-output = 1.20
-
-[limit]
-context = 163_840
-output = 163_840
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/chutes/models/google/gemma-4-31B-turbo-TEE.toml

@@ -0,0 +1,25 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
+name = "gemma 4 31B turbo TEE"
+family = "gemma"
+release_date = "2026-04-25"
+last_updated = "2026-04-25"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[cost]
+input = 0.13
+output = 0.38
+cache_read = 0.065
+
+[limit]
+context = 131_072
+output = 65_536
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

providers/chutes/models/miromind-ai/MiroThinker-v1.5-235B.toml

@@ -1,22 +0,0 @@
-name = "MiroThinker V1.5 235B"
-release_date = "2026-01-10"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = false
-structured_output = false
-open_weights = true
-
-[cost]
-input = 0.30
-output = 1.20
-cache_read = 0.15
-
-[limit]
-context = 262_144
-output = 8_192
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/chutes/models/moonshotai/Kimi-K2-Instruct-0905.toml

@@ -1,23 +0,0 @@
-name = "Kimi K2 Instruct 0905"
-family = "kimi"
-release_date = "2025-12-29"
-last_updated = "2026-01-10"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = true
-
-[cost]
-input = 0.39
-output = 1.90
-cache_read = 0.195
-
-[limit]
-context = 262_144
-output = 262_144
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/chutes/models/moonshotai/Kimi-K2.5-TEE.toml

@@ -1,21 +1,21 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Kimi K2.5 TEE"
 family = "kimi"
 release_date = "2026-01-27"
-last_updated = "2026-01-27"
-attachment = false
+last_updated = "2026-04-25"
+attachment = true
 reasoning = true
 temperature = true
 tool_call = true
-knowledge = "2024-10"
-open_weights = true
 structured_output = true
-
-[interleaved]
-field = "reasoning_content"
+open_weights = true
+knowledge = "2024-10"
 
 [cost]
-input = 0.60
-output = 3.00
+input = 0.44
+output = 2
+cache_read = 0.22
 
 [limit]
 context = 262_144
@@ -24,3 +24,6 @@ output = 65_535
 [modalities]
 input = ["text", "image", "video"]
 output = ["text"]
+
+[interleaved]
+field = "reasoning_content"

providers/chutes/models/moonshotai/Kimi-K2.6-TEE.toml

@@ -1,26 +1,29 @@
+# Auto-generated by generate-chutes.ts — do not edit pricing, limits, or capabilities.
+# Manual overrides preserved on re-run: name, family, knowledge, interleaved, status
 name = "Kimi K2.6 TEE"
 family = "kimi"
 release_date = "2026-04-20"
-last_updated = "2026-04-23"
+last_updated = "2026-04-25"
 attachment = true
 reasoning = true
 temperature = true
 tool_call = true
-knowledge = "2025-12"
-open_weights = true
 structured_output = true
-
-[interleaved]
-field = "reasoning_content"
+open_weights = true
+knowledge = "2025-12"
 
 [cost]
-input = 0.44
-output = 2.00
+input = 0.95
+output = 4
+cache_read = 0.475
 
 [limit]
 context = 262_144
-output = 262_144
+output = 65_535
 
 [modalities]
 input = ["text", "image", "video"]
 output = ["text"]
+
+[interleaved]
+field = "reasoning_content"