fix: [Price] Wrong price for google/gemini-3.6-flash, for Google provider

models/deepseek/deepseek-v4-flash-vision-exp.toml

@@ -1,19 +0,0 @@
-name = "DeepSeek V4 Flash Vision Exp"
-description = "Experimental multimodal DeepSeek V4 Flash model for image understanding, coding, and agentic work"
-family = "deepseek-flash"
-release_date = "2026-08-21"
-last_updated = "2026-08-21"
-attachment = true
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = false
-
-[limit]
-context = 1_000_000
-output = 384_000
-
-[modalities]
-input = ["text", "image"]
-output = ["text"]

models/deepseek/deepseek-v4-pro-0423.toml

@@ -1,22 +0,0 @@
-# https://ofox.ai/models/deepseek/deepseek-v4-pro-0423
-# https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro
-name = "DeepSeek V4 Pro 0423"
-description = "DeepSeek V4 Pro initial snapshot with million-token context and support for thinking and non-thinking modes"
-family = "deepseek-thinking"
-release_date = "2026-04-23"
-last_updated = "2026-04-23"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-knowledge = "2025-05"
-open_weights = true
-
-[limit]
-context = 1_000_000
-output = 384_000
-
-[modalities]
-input = ["text"]
-output = ["text"]

models/deepseek/deepseek-v4-pro-0813.toml

@@ -1,16 +1,14 @@
-# https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813
 name = "DeepSeek V4 Pro 0813"
 description = "DeepSeek V4 Pro snapshot with million-token context and support for thinking and non-thinking modes"
 family = "deepseek-thinking"
 release_date = "2026-08-12"
-last_updated = "2026-08-22"
+last_updated = "2026-08-12"
 attachment = false
 reasoning = true
 temperature = true
 tool_call = true
 structured_output = true
-open_weights = true
-license = "MIT"
+open_weights = false
 
 [limit]
 context = 1_000_000
@@ -19,7 +17,3 @@ output = 384_000
 [modalities]
 input = ["text"]
 output = ["text"]
-
-[[weights]]
-label = "Hugging Face"
-url = "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813"

models/google/gemini-flash-latest.toml

@@ -1,15 +1,15 @@
-# Tracks the current Gemini Flash release (gemini-3.7-flash).
+# Tracks the current Gemini Flash release (gemini-3.5-flash).
 name = "Gemini Flash Latest"
-description = "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning"
+description = "Fast Gemini model balancing multimodal reasoning, tool use, and cost"
 family = "gemini-flash"
-release_date = "2026-08-13"
-last_updated = "2026-08-13"
+release_date = "2026-05-19"
+last_updated = "2026-05-19"
 attachment = true
 reasoning = true
 temperature = true
 tool_call = true
 structured_output = true
-knowledge = "2026-03"
+knowledge = "2025-01"
 open_weights = false
 
 [limit]

models/google/gemini-flash-lite-latest.toml

@@ -1,15 +1,15 @@
-# Tracks the current Gemini Flash-Lite release (gemini-3.5-flash-lite).
+# Tracks the current Gemini Flash-Lite release (gemini-3.1-flash-lite).
 name = "Gemini Flash-Lite Latest"
-description = "Fast Gemini model balancing multimodal reasoning, tool use, and cost"
+description = "Low-latency Gemini model for high-volume multimodal and agent workloads"
 family = "gemini-flash-lite"
-release_date = "2026-07-21"
-last_updated = "2026-07-21"
+release_date = "2026-05-07"
+last_updated = "2026-05-07"
 attachment = true
 reasoning = true
 temperature = true
 tool_call = true
 structured_output = true
-knowledge = "2026-03"
+knowledge = "2025-01"
 open_weights = false
 
 [limit]

packages/core/src/sync/auto-merge.ts

@@ -11,7 +11,6 @@ const REVIEWED_REASONING_PROVIDERS = new Set([
   "hyper",
   "kilo",
   "llmgateway",
-  "llmgateway-providers",
   "merge-gateway",
   "nano-gpt",
   "openrouter",

packages/core/test/auto-merge.test.ts

@@ -82,7 +82,7 @@ test("does not inspect deleted models", async () => {
 });
 
 test("allows reviewed providers with explicit reasoning options", async () => {
-  for (const provider of ["crossmodel", "edenai", "empiriolabs", "hyper", "kilo", "llmgateway", "llmgateway-providers", "merge-gateway", "nano-gpt", "openrouter", "venice"]) {
+  for (const provider of ["crossmodel", "edenai", "empiriolabs", "hyper", "kilo", "llmgateway", "merge-gateway", "nano-gpt", "openrouter", "venice"]) {
     const decision = await classifyAutoMerge(
       [{ status: "updated", path: `providers/${provider}/models/reasoner.toml` }],
       async () => fullModel(true, 'reasoning_options = [{ type = "toggle" }]'),

packages/sdk/src/types.ts

@@ -68,10 +68,7 @@ export interface CostTier extends Cost {
 
 /** Pricing for a provider's model, including context-size tiers. */
 export interface ModelCost extends Cost {
-  /**
-   * Legacy compatibility field for context-tier pricing.
-   * @deprecated Use `tiers` to read the exact context threshold.
-   */
+  /** Legacy compatibility field: pricing applied beyond 200K context. Prefer `tiers`. */
   context_over_200k?: Cost
   /** Context-size-based pricing tiers. */
   tiers?: CostTier[]

providers/aki-io/models/deepseek-v4-flash-0731-284b.toml

@@ -1,13 +0,0 @@
-# attachment are not supported on AKI.IO for security reasons and therefore are disabled
-base_model = "deepseek/deepseek-v4-flash-0731"
-reasoning_options = [{ type = "effort", values = ["none", "low", "high", "max"] }]
-
-# cost as described on the AKI.IO pricing page at https://aki.io/#pricing
-[cost]
-input = 0.20
-output = 0.50
-cache_read = 0.10
-
-[limit]
-context = 1_048_576
-output = 81_920

providers/aki-io/models/gemma4-26b.toml

@@ -2,6 +2,7 @@
 # AKI.IO uses the reasoning_option 'none' to disable reasoning
 base_model = "google/gemma-4-26b-a4b-it"
 attachment = false
+reasoning = true
 reasoning_options = [{ type = "effort", values = ["none", "high"] }]
 
 [cost]

providers/aki-io/models/kimi-k2.7-code-1100b.toml

@@ -4,11 +4,9 @@ base_model = "moonshotai/kimi-k2.7-code"
 attachment = false
 reasoning_options = []
 
-# cost as described on the AKI.IO pricing page at https://aki.io/#pricing
 [cost]
 input = 0.86
 output = 3.00
-cache_read = 0.18
 
 [limit]
 context = 262_144

providers/aki-io/models/minimax-m2.5-230b.toml

@@ -0,0 +1,12 @@
+# attachment are not supported on AKI.IO for security reasons and therefore are disabled
+base_model = "minimax/MiniMax-M2.5"
+attachment = false 
+reasoning_options = []
+
+[cost]
+input = 0.25
+output = 1.20
+
+[limit]
+context = 196_608
+output = 65_536

providers/aki-io/models/qwen3.8-27b.toml

@@ -1,14 +0,0 @@
-# attachment are not supported on AKI.IO for security reasons and therefore are disabled
-# AKI.IO uses the reasoning_option 'none' to disable reasoning
-base_model = "alibaba/qwen3.8-27b"
-attachment = false
-reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "xhigh"] }]
-
-# cost as described on the AKI.IO pricing page at https://aki.io/#pricing
-[cost]
-input = 0.30
-output = 2.20
-cache_read = 0.10
-
-[modalities]
-input = ["text", "image"]

providers/crof/models/glm-4.7-flash.toml

@@ -0,0 +1,16 @@
+base_model = "zhipuai/glm-4.7-flash"
+# Crof's request docs establish no model-specific toggle, effort, or budget.
+# https://crof.ai/docs.md (accessed 2026-06-25)
+reasoning_options = []
+
+[cost]
+input = 0.04
+output = 0.3
+cache_read = 0.008
+cache_write = 0
+
+[limit]
+context = 202_752
+
+[provider]
+npm = "@ai-sdk/openai-compatible"

providers/crof/models/glm-4.7.toml

@@ -0,0 +1,20 @@
+base_model = "zhipuai/glm-4.7"
+# Crof's request docs establish no model-specific toggle, effort, or budget.
+# https://crof.ai/docs.md (accessed 2026-06-25)
+reasoning_options = []
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 0.25
+output = 1.1
+cache_read = 0.05
+cache_write = 0
+
+[limit]
+context = 202_752
+output = 202_752
+
+[provider]
+npm = "@ai-sdk/openai-compatible"

providers/crof/models/glm-5.toml

@@ -0,0 +1,20 @@
+base_model = "zhipuai/glm-5"
+# Crof's request docs establish no model-specific toggle, effort, or budget.
+# https://crof.ai/docs.md (accessed 2026-06-25)
+reasoning_options = []
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 0.48
+output = 1.9
+cache_read = 0.1
+cache_write = 0
+
+[limit]
+context = 202_752
+output = 202_752
+
+[provider]
+npm = "@ai-sdk/openai-compatible"

providers/crof/models/kimi-k2.5-lightning.toml

@@ -0,0 +1,28 @@
+name = "Kimi K2.5 (Lightning)"
+description = "Kimi multimodal agent model for visual understanding, coding, and planning"
+family = "kimi-k2"
+release_date = "2026-02-06"
+last_updated = "2026-02-06"
+attachment = true
+reasoning = true
+reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "high"] }]
+temperature = false
+tool_call = true
+structured_output = true
+open_weights = true
+
+[interleaved]
+field = "reasoning_content"
+
+[cost]
+input = 1.00
+output = 3.00
+cache_read = 0.20
+
+[limit]
+context = 131_072
+output = 32_768
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]

providers/crof/models/kimi-k2.5.toml

@@ -1,13 +1,13 @@
-base_model = "alibaba/qwen3.8-27b"
+base_model = "moonshotai/kimi-k2.5"
 reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "high"] }]
 
-[cost]
-input = 0.25
-output = 2.1
-cache_read = 0.06
+[interleaved]
+field = "reasoning_content"
 
-[limit]
-output = 262_144
+[cost]
+input = 0.35
+output = 1.7
+cache_read = 0.07
 
 [provider]
 npm = "@ai-sdk/openai-compatible"

providers/crof/models/minimax-m2.5.toml

@@ -0,0 +1,14 @@
+base_model = "minimax/MiniMax-M2.5"
+# Crof's request docs establish no model-specific toggle, effort, or budget.
+# https://crof.ai/docs.md (accessed 2026-06-25)
+reasoning = true
+reasoning_options = []
+
+[cost]
+input = 0.11
+output = 0.95
+cache_read = 0.02
+cache_write = 0.375
+
+[provider]
+npm = "@ai-sdk/openai-compatible"

providers/crossmodel/models/deepseek/deepseek-v4-flash-vision-exp.toml

@@ -1,14 +0,0 @@
-base_model = "deepseek/deepseek-v4-flash-vision-exp"
-
-[[reasoning_options]]
-type = "toggle"
-
-[[reasoning_options]]
-type = "effort"
-values = ["low", "medium", "high", "xhigh", "max"]
-
-[cost]
-input = 0.405
-output = 1.215
-cache_read = 0.0135
-cache_write = 0.405

providers/deepseek/models/deepseek-v4-flash-vision-exp.toml

@@ -1,23 +0,0 @@
-# DeepSeek-V4-Flash-Vision-Exp is priced the same as DeepSeek V4 Flash.
-# https://api-docs.deepseek.com/quick_start/pricing/ (accessed 2026-08-22)
-# https://api-docs.deepseek.com/guides/vision (accessed 2026-08-22)
-# OpenAI: `thinking.type = enabled|disabled`, `reasoning_effort = low|high|max`.
-# Anthropic: `thinking.type`, `output_config.effort = low|high|max`; budget ignored.
-base_model = "deepseek/deepseek-v4-flash-vision-exp"
-status = "beta"
-
-[[reasoning_options]]
-type = "toggle"
-
-[[reasoning_options]]
-type = "effort"
-values = ["low", "high", "max"]
-
-[interleaved]
-field = "reasoning_content"
-
-[cost]
-input = 0.14
-output = 0.28
-reasoning = 0.28
-cache_read = 0.0028

providers/digitalocean/models/qwen3.8-max.toml

@@ -11,6 +11,7 @@ output = 6
 cache_read = 0.2
 
 [limit]
+context = 262_144
 output = 262_144
 
 [modalities]

providers/edenai/models/deepinfra/deepseek-ai/DeepSeek-V4-Pro-0813.toml

@@ -1,4 +1,6 @@
 base_model = "deepseek/deepseek-v4-pro-0813"
+tool_call = false
+structured_output = false
 
 [[reasoning_options]]
 type = "effort"

providers/edenai/models/deepseek/deepseek-v4-flash-vision-exp.toml

@@ -1,13 +0,0 @@
-base_model = "deepseek/deepseek-v4-flash-vision-exp"
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "low", "high", "max"]
-
-[cost]
-input = 0.22
-output = 0.66
-cache_read = 0.007
-
-[limit]
-context = 1_048_576

providers/edenai/models/deepseek/deepseek-v4-flash.toml

@@ -5,6 +5,9 @@ type = "effort"
 values = ["none", "low", "high", "max"]
 
 [cost]
-input = 0.44
-output = 1.32
-cache_read = 0.014
+input = 0.22
+output = 0.66
+cache_read = 0.007
+
+[limit]
+context = 1_048_576

providers/edenai/models/deepseek/deepseek-v4-pro.toml

@@ -5,6 +5,9 @@ type = "effort"
 values = ["none", "high", "max"]
 
 [cost]
-input = 1.32
-output = 3.96
-cache_read = 0.044
+input = 0.66
+output = 1.98
+cache_read = 0.022
+
+[limit]
+context = 1_048_576

providers/edenai/models/ionos/meta-llama/Llama-3.3-70B-Instruct.toml

@@ -4,5 +4,5 @@ tool_call = false
 structured_output = false
 
 [cost]
-input = 0.760435
-output = 0.760435
+input = 0.759265
+output = 0.759265

providers/edenai/models/ionos/openai/gpt-oss-120b.toml

@@ -7,5 +7,5 @@ type = "effort"
 values = ["low", "medium", "high"]
 
 [cost]
-input = 0.175485
-output = 0.760435
+input = 0.175215
+output = 0.759265

providers/edenai/models/nebius/deepseek-ai/DeepSeek-V4-Flash-0731.toml

@@ -1,13 +0,0 @@
-base_model = "deepseek/deepseek-v4-flash-0731"
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "low", "high", "max"]
-
-[cost]
-input = 0.14
-output = 0.28
-cache_read = 0.14
-
-[limit]
-context = 1_048_576

providers/edenai/models/nebius/nvidia/nemotron-3-super-120b-a12b.toml

@@ -9,3 +9,6 @@ values = ["none", "low", "medium"]
 input = 0.3
 output = 0.9
 cache_read = 0.3
+
+[limit]
+context = 8_000

providers/edenai/models/openai/gpt-5.6-sol.toml

@@ -5,14 +5,14 @@ type = "effort"
 values = ["none", "low", "medium", "high", "xhigh", "max"]
 
 [cost]
-input = 4
-output = 20
-cache_read = 0.4
-cache_write = 5
+input = 5
+output = 30
+cache_read = 0.5
+cache_write = 6.25
 
 [[cost.tiers]]
 tier = { type = "context", size = 272_000 }
-input = 8
-output = 30
-cache_read = 0.8
-cache_write = 10
+input = 10
+output = 45
+cache_read = 1
+cache_write = 12.5

providers/edenai/models/openai/gpt-latest.toml

@@ -5,14 +5,14 @@ type = "effort"
 values = ["none", "low", "medium", "high", "xhigh", "max"]
 
 [cost]
-input = 4
-output = 20
-cache_read = 0.4
-cache_write = 5
+input = 5
+output = 30
+cache_read = 0.5
+cache_write = 6.25
 
 [[cost.tiers]]
 tier = { type = "context", size = 272_000 }
-input = 8
-output = 30
-cache_read = 0.8
-cache_write = 10
+input = 10
+output = 45
+cache_read = 1
+cache_write = 12.5

providers/edenai/models/qwen/deepseek-v4-flash-0731.toml

@@ -5,5 +5,5 @@ type = "effort"
 values = ["none", "low", "high", "max"]
 
 [cost]
-input = 0.176
-output = 0.528
+input = 0.40172
+output = 1.20516

providers/edenai/models/qwen/deepseek-v4-pro-0813.toml

@@ -5,5 +5,5 @@ type = "effort"
 values = ["none", "low", "high", "max"]
 
 [cost]
-input = 0.627
-output = 1.881
+input = 1.20516
+output = 3.61548

providers/edenai/models/scaleway/deepseek-v4-flash-0731.toml

@@ -6,8 +6,8 @@ type = "effort"
 values = ["none", "low", "high", "max"]
 
 [cost]
-input = 0.46796
-output = 0.93592
+input = 0.46724
+output = 0.93448
 
 [limit]
 context = 256_000

providers/edenai/models/scaleway/gpt-oss-120b.toml

@@ -6,8 +6,8 @@ type = "effort"
 values = ["low", "medium", "high"]
 
 [cost]
-input = 0.175485
-output = 0.70194
+input = 0.175215
+output = 0.70086
 
 [limit]
 context = 128_000

providers/edenai/models/scaleway/llama-3.3-70b-instruct.toml

@@ -3,5 +3,5 @@ attachment = false
 structured_output = false
 
 [cost]
-input = 1.05291
-output = 1.05291
+input = 1.05129
+output = 1.05129

providers/empiriolabs/models/deepseek-v4-flash-0731.toml

@@ -3,6 +3,10 @@ base_model = "deepseek/deepseek-v4-flash-0731"
 [[reasoning_options]]
 type = "toggle"
 
+[[reasoning_options]]
+type = "effort"
+values = ["none", "low", "medium", "high", "max"]
+
 [cost]
 input = 0.424
 output = 1.272

providers/empiriolabs/models/deepseek-v4-flash.toml

@@ -4,6 +4,10 @@ last_updated = "2026-06-12"
 [[reasoning_options]]
 type = "toggle"
 
+[[reasoning_options]]
+type = "effort"
+values = ["none", "low", "medium", "high", "max"]
+
 [[reasoning_options]]
 type = "budget_tokens"
 min = 1

providers/empiriolabs/models/deepseek-v4-pro-0813.toml

@@ -7,11 +7,6 @@ type = "toggle"
 type = "effort"
 values = ["none", "low", "medium", "high", "max"]
 
-[[reasoning_options]]
-type = "budget_tokens"
-min = 1
-max = 393_216
-
 [cost]
 input = 1.32
 output = 3.96

providers/empiriolabs/models/deepseek-v4-pro.toml

@@ -4,6 +4,10 @@ last_updated = "2026-06-12"
 [[reasoning_options]]
 type = "toggle"
 
+[[reasoning_options]]
+type = "effort"
+values = ["none", "low", "medium", "high", "max"]
+
 [[reasoning_options]]
 type = "budget_tokens"
 min = 1

providers/empiriolabs/models/glm-5-1.toml

@@ -5,6 +5,10 @@ last_updated = "2026-06-12"
 [[reasoning_options]]
 type = "toggle"
 
+[[reasoning_options]]
+type = "effort"
+values = ["none", "low", "medium", "high", "max"]
+
 [[reasoning_options]]
 type = "budget_tokens"
 min = 1

providers/empiriolabs/models/kimi-k2-6.toml

@@ -4,6 +4,15 @@ last_updated = "2026-06-12"
 [[reasoning_options]]
 type = "toggle"
 
+[[reasoning_options]]
+type = "effort"
+values = ["none", "low", "medium", "high", "max"]
+
+[[reasoning_options]]
+type = "budget_tokens"
+min = 1
+max = 81_920
+
 [cost]
 input = 0.8939
 output = 3.7131

providers/github-copilot/models/gpt-5.6-sol.toml

@@ -1,19 +1,15 @@
-# Source: https://docs.github.com/en/copilot/reference/copilot-billing/models-and-pricing
-# GPT-5.6 Sol promotional pricing (50% off standard rates) through September 3, 2026.
-# Default ≤272K: $2.50 / $0.25 / $3.125 / $15.00 (input / cache_read / cache_write / output)
-# Long context >272K: $5.00 / $0.50 / $6.25 / $22.50
 base_model = "openai/gpt-5.6-sol"
 reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "high", "xhigh", "max"] }]
 
 [cost]
-input = 2.5
-output = 15
-cache_read = 0.25
-cache_write = 3.125
+input = 5
+output = 30
+cache_read = 0.5
+cache_write = 6.25
 
 [[cost.tiers]]
 tier = { type = "context", size = 272_000 }
-input = 5
-output = 22.5
-cache_read = 0.5
-cache_write = 6.25
+input = 10
+output = 45
+cache_read = 1
+cache_write = 12.5

providers/google-vertex-anthropic/models/claude-fable-5@default.toml

@@ -1,9 +0,0 @@
-# GA on Vertex AI Model Garden: publishers/anthropic/models/claude-fable-5 (versionId: default)
-base_model = "anthropic/claude-fable-5"
-reasoning_options = [{ type = "effort", values = ["low", "medium", "high", "xhigh", "max"] }]
-
-[cost]
-input = 10
-output = 50
-cache_read = 1
-cache_write = 12.5

providers/google-vertex/models/claude-fable-5@default.toml

@@ -1,12 +0,0 @@
-# GA on Vertex AI Model Garden: publishers/anthropic/models/claude-fable-5 (versionId: default)
-base_model = "anthropic/claude-fable-5"
-reasoning_options = [{ type = "effort", values = ["low", "medium", "high", "xhigh", "max"] }]
-
-[cost]
-input = 10
-output = 50
-cache_read = 1
-cache_write = 12.5
-
-[provider]
-npm = "@ai-sdk/google-vertex/anthropic"

providers/google/models/gemini-3.6-flash.toml

@@ -1,7 +1,8 @@
 # Sources:
 # - https://ai.google.dev/gemini-api/docs/pricing#gemini-3.6-flash
-# Introductory Standard pricing applies through 2026-12-31. The published
-# 2027 rates are $1.50 input, $7.50 output, and $0.15 cache read.
+# Introductory Standard paid tier through 2026-12-31: $0.75/$3.75 (input/output);
+# $0.075 cache read. Reverts to $1.50/$7.50 and $0.15 cache from 2027-01-01.
+# Input (text/image/video/audio) shares the input rate on the pricing page.
 base_model = "google/gemini-3.6-flash"
 
 [[reasoning_options]]

providers/google/models/gemini-flash-latest.toml

@@ -1,27 +1,27 @@
-# Alias for the current Gemini Flash release (gemini-3.7-flash).
-# Keep reasoning_options and cost in sync with gemini-3.7-flash.
+# Alias for the current Gemini Flash release (gemini-3.5-flash).
+# Keep reasoning_options and cost in sync with gemini-3.5-flash.
 name = "Gemini Flash Latest"
-description = "High-efficiency Gemini model for agentic workflows, coding, and multimodal reasoning"
+description = "Fast Gemini model balancing multimodal reasoning, tool use, and cost"
 family = "gemini-flash"
-release_date = "2026-08-13"
-last_updated = "2026-08-13"
+release_date = "2026-05-19"
+last_updated = "2026-05-19"
 attachment = true
 reasoning = true
 temperature = true
 tool_call = true
 structured_output = true
-knowledge = "2026-03"
+knowledge = "2025-01"
 open_weights = false
 
 [[reasoning_options]]
 type = "effort"
-values = ["low", "medium", "high"]
+values = ["minimal", "low", "medium", "high"]
 
 [cost]
-input = 0.75
-output = 3.75
-cache_read = 0.075
-input_audio = 0.75
+input = 1.50
+output = 9.00
+cache_read = 0.15
+input_audio = 1.50
 
 [limit]
 context = 1_048_576

providers/google/models/gemini-flash-lite-latest.toml

@@ -1,16 +1,16 @@
-# Alias for the current Gemini Flash-Lite release (gemini-3.5-flash-lite).
-# Keep reasoning_options and cost in sync with gemini-3.5-flash-lite.
+# Alias for the current Gemini Flash-Lite release (gemini-3.1-flash-lite).
+# Keep reasoning_options and cost in sync with gemini-3.1-flash-lite.
 name = "Gemini Flash-Lite Latest"
-description = "Fast Gemini model balancing multimodal reasoning, tool use, and cost"
+description = "Low-latency Gemini model for high-volume multimodal and agent workloads"
 family = "gemini-flash-lite"
-release_date = "2026-07-21"
-last_updated = "2026-07-21"
+release_date = "2026-05-07"
+last_updated = "2026-05-07"
 attachment = true
 reasoning = true
 temperature = true
 tool_call = true
 structured_output = true
-knowledge = "2026-03"
+knowledge = "2025-01"
 open_weights = false
 
 [[reasoning_options]]
@@ -18,9 +18,10 @@ type = "effort"
 values = ["minimal", "low", "medium", "high"]
 
 [cost]
-input = 0.30
-output = 2.50
-cache_read = 0.03
+input = 0.25
+output = 1.50
+cache_read = 0.025
+input_audio = 0.50
 
 [limit]
 context = 1_048_576

providers/hyper/models/gemma-4-26b-a4b-it.toml

@@ -5,9 +5,9 @@ attachment = false
 reasoning_options = []
 
 [cost]
-input = 0.11
-output = 0.408
-cache_write = 0.055
+input = 0.12
+output = 0.42
+cache_write = 0.06
 
 [limit]
 context = 256_000

providers/hyper/models/glm-5.toml

@@ -4,9 +4,9 @@ last_updated = "2026-08-16"
 reasoning_options = []
 
 [cost]
-input = 0.91
-output = 2.934
-cache_write = 0.455
+input = 0.83
+output = 2.558
+cache_write = 0.415
 
 [limit]
 context = 202_752

providers/hyper/models/kimi-k2.5.toml

@@ -5,9 +5,9 @@ attachment = false
 reasoning_options = []
 
 [cost]
-input = 0.5444
-output = 2.855
-cache_write = 0.2722
+input = 0.5504
+output = 2.885
+cache_write = 0.2752
 
 [limit]
 output = 26_214

providers/hyper/models/kimi-k3.toml

@@ -11,8 +11,5 @@ input = 3.2664
 output = 16.332
 cache_read = 0.32664
 
-[limit]
-output = 16_000
-
 [modalities]
 input = ["text", "image"]

providers/kilo/models/deepcogito/cogito-v2.1-671b.toml

@@ -0,0 +1,27 @@
+name = "Deep Cogito: Cogito v2.1 671B"
+description = "Reasoning model for deliberate analysis, multi-step problem solving, and tool use"
+family = "cogito"
+release_date = "2025-11-13"
+last_updated = "2025-11-13"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = false
+structured_output = true
+open_weights = false
+
+[[reasoning_options]]
+type = "effort"
+values = ["none", "high"]
+
+[cost]
+input = 1.25
+output = 1.25
+
+[limit]
+context = 128_000
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]

providers/kilo/models/deepseek/deepseek-chat-v3.1.toml

@@ -20,8 +20,8 @@ output = 1
 cache_read = 0.135
 
 [limit]
-context = 161_000
-output = 161_000
+context = 163_840
+output = 32_768
 
 [modalities]
 input = ["text"]

providers/kilo/models/deepseek/deepseek-v3.2.toml

@@ -12,4 +12,4 @@ cache_read = 0.1345
 
 [limit]
 context = 163_840
-output = 163_840
+output = 65_536

providers/kilo/models/deepseek/deepseek-v4-flash-0731.toml

@@ -6,9 +6,10 @@ type = "effort"
 values = ["none", "low", "high", "max"]
 
 [cost]
-input = 0.44
-output = 1.32
+input = 0.14
+output = 0.28
 cache_read = 0.028
 
 [limit]
 context = 1_048_576
+output = 393_216

providers/kilo/models/deepseek/deepseek-v4-flash-vision-exp.toml

@@ -1,15 +0,0 @@
-base_model = "deepseek/deepseek-v4-flash-vision-exp"
-description = "DeepSeek V4 Flash Vision Exp is an experimental vision-enabled version of [DeepSeek V4 Flash 0731](https://openrouter.ai/deepseek/deepseek-v4-flash-0731) from DeepSeek, adding image understanding while matching the base model on text capabilities including agents,..."
-structured_output = false
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "low", "high", "max"]
-
-[cost]
-input = 0.22
-output = 0.66
-cache_read = 0.007
-
-[limit]
-context = 1_048_576

providers/kilo/models/deepseek/deepseek-v4-pro.toml

@@ -11,4 +11,5 @@ output = 3.2
 cache_read = 0.135
 
 [limit]
-context = 1_024_000
+context = 1_048_576
+output = 393_216

providers/kilo/models/google/gemma-4-26b-a4b-it.toml

@@ -6,8 +6,8 @@ type = "effort"
 values = ["none", "high"]
 
 [cost]
-input = 0.042
-output = 0.22
+input = 0.05
+output = 0.25
 
 [limit]
 output = 16_384

providers/kilo/models/meituan/longcat-2.0-free.toml

@@ -1,29 +0,0 @@
-name = "Meituan: LongCat 2.0 (free)"
-description = "LongCat 2.0 is a sparse mixture-of-experts language model from Meituan, with 48B active parameters out of 1.6T total. It is suited for coding, repository-level changes, long-horizon problem solving, and agentic workflows. Available free in Kilo for a limited time."
-family = "longcat"
-release_date = "2025-08-26"
-last_updated = "2025-08-26"
-attachment = false
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = false
-open_weights = false
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "high"]
-
-[cost]
-input = 0
-output = 0
-reasoning = 0
-cache_read = 0
-
-[limit]
-context = 1_048_756
-output = 131_072
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/kilo/models/meta/muse-spark-1.2-contributor.toml

@@ -1,28 +0,0 @@
-name = "Meta: Muse Spark 1.2 Contributor"
-description = "Muse Spark 1.2 contributor tier is a reasoning model from Meta designed for developers who want to start building at an even lower cost. It’s meaningfully cheaper than Muse Spark..."
-family = "muse"
-release_date = "2026-08-21"
-last_updated = "2026-08-21"
-attachment = true
-reasoning = true
-temperature = true
-tool_call = true
-structured_output = true
-open_weights = false
-
-[[reasoning_options]]
-type = "effort"
-values = ["minimal", "low", "medium", "high", "xhigh"]
-
-[cost]
-input = 0.1
-output = 0.2
-cache_read = 0.002
-
-[limit]
-context = 1_048_576
-output = 1_048_576
-
-[modalities]
-input = ["text", "image", "video", "pdf", "audio"]
-output = ["text"]

providers/kilo/models/minimax/minimax-m2.5.toml

@@ -12,5 +12,5 @@ output = 1.2
 cache_read = 0.03
 
 [limit]
-context = 200_000
-output = 128_000
+context = 198_000
+output = 32_768

providers/kilo/models/minimax/minimax-m2.7.toml

@@ -10,6 +10,3 @@ values = ["high"]
 input = 0.3
 output = 1.2
 cache_read = 0.06
-
-[limit]
-context = 196_608

providers/kilo/models/mistralai/mistral-small-3.2-24b-instruct.toml

@@ -11,11 +11,11 @@ structured_output = true
 open_weights = false
 
 [cost]
-input = 0.075
-output = 0.2
+input = 0.09375
+output = 0.25
 
 [limit]
-context = 128_000
+context = 256_000
 output = 16_384
 
 [modalities]

providers/kilo/models/openai/gpt-5.6-sol-pro.toml

@@ -6,7 +6,7 @@ type = "effort"
 values = ["none", "low", "medium", "high", "xhigh", "max"]
 
 [cost]
-input = 4
-output = 20
-cache_read = 0.4
-cache_write = 5
+input = 5
+output = 30
+cache_read = 0.5
+cache_write = 6.25

providers/kilo/models/openai/gpt-5.6-sol.toml

@@ -6,7 +6,7 @@ type = "effort"
 values = ["none", "low", "medium", "high", "xhigh", "max"]
 
 [cost]
-input = 4
-output = 20
-cache_read = 0.4
-cache_write = 5
+input = 5
+output = 30
+cache_read = 0.5
+cache_write = 6.25

providers/kilo/models/qwen/qwen3-14b.toml

@@ -19,8 +19,8 @@ input = 0.2275
 output = 0.91
 
 [limit]
-context = 40_960
-output = 16_384
+context = 131_072
+output = 8_192
 
 [modalities]
 input = ["text"]

providers/kilo/models/qwen/qwen3-next-80b-a3b-instruct.toml

@@ -8,4 +8,4 @@ output = 0.78
 
 [limit]
 context = 262_144
-output = 262_144
+output = 16_384

providers/kilo/models/qwen/qwen3.8-2.4t-a95b.toml

@@ -2,6 +2,7 @@
 # https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B
 # https://openrouter.ai/qwen/qwen3.8-2.4t-a95b
 # Kilo host pricing retained from prior catalog entry; lab metadata via base_model.
+
 base_model = "alibaba/qwen3.8-2.4t-a95b"
 
 [[reasoning_options]]
@@ -15,4 +16,5 @@ cache_read = 0.25
 cache_write = 2.5
 
 [limit]
-context = 1_048_576
+context = 1_000_000
+output = 262_144

providers/kilo/models/qwen/qwen3.8-27b.toml

@@ -6,10 +6,10 @@ type = "effort"
 values = ["none", "low", "medium", "xhigh"]
 
 [cost]
-input = 0.5
-output = 3
-cache_read = 0.1
-cache_write = 0.625
+input = 0.575
+output = 3.45
+cache_read = 0.115
+cache_write = 0.71875
 
 [limit]
 output = 131_072

providers/kilo/models/tencent/hy-mt2-7b.toml

@@ -1,23 +0,0 @@
-name = "Tencent: Hy-MT2-7B"
-description = "Hy-MT2-7B is a 7B-parameter translation model from Tencent. It supports 33 language pairs and five Chinese dialect and minority-language pairs, with workflows for structured, delimiter-based, contextual, glossary-based, and style-guided translation."
-family = "Hy"
-release_date = "2026-08-19"
-last_updated = "2026-08-19"
-attachment = false
-reasoning = false
-temperature = true
-tool_call = false
-structured_output = true
-open_weights = false
-
-[cost]
-input = 0.074
-output = 0.295
-
-[limit]
-context = 8_192
-output = 4_096
-
-[modalities]
-input = ["text"]
-output = ["text"]

providers/kilo/models/tencent/hy3.toml

@@ -7,9 +7,9 @@ type = "effort"
 values = ["none", "low", "high"]
 
 [cost]
-input = 0.132
-output = 0.528
-cache_read = 0.033
+input = 0.14
+output = 0.58
+cache_read = 0.035
 
 [limit]
 context = 262_144

providers/kilo/models/thinkingmachines/inkling-small.toml

@@ -1,6 +1,6 @@
 base_model = "thinkingmachines/inkling-small"
 description = "Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of..."
-structured_output = false
+structured_output = true
 
 [[reasoning_options]]
 type = "effort"

providers/kilo/models/thinkingmachines/inkling-small:free.toml

@@ -1,16 +0,0 @@
-base_model = "thinkingmachines/inkling-small"
-name = "Thinking Machines: Inkling Small (free)"
-description = "Inkling Small is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 12B active parameters out of 276B total. It is positioned as the smaller, more efficient member of..."
-structured_output = false
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "minimal", "low", "medium", "high", "max"]
-
-[cost]
-input = 0
-output = 0
-
-[limit]
-context = 262_144
-output = 262_144

providers/kilo/models/thinkingmachines/inkling:free.toml

@@ -1,16 +0,0 @@
-base_model = "thinkingmachines/inkling"
-name = "Thinking Machines: Inkling (free)"
-description = "Inkling is an open-weight multimodal mixture-of-experts model from Thinking Machines Lab, with 41B active parameters out of 975B total. It is designed for general-purpose reasoning, coding, agentic and tool-use systems,..."
-structured_output = false
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "minimal", "low", "medium", "high", "max"]
-
-[cost]
-input = 0
-output = 0
-
-[limit]
-context = 262_144
-output = 262_144

providers/kilo/models/~deepseek/deepseek-v4-flash-latest.toml

@@ -15,13 +15,13 @@ type = "effort"
 values = ["none", "low", "high", "max"]
 
 [cost]
-input = 0.04
-output = 0.08
-cache_read = 0.008
+input = 0.065
+output = 0.14
+cache_read = 0.014
 
 [limit]
-context = 1_048_576
-output = 1_048_576
+context = 262_144
+output = 262_144
 
 [modalities]
 input = ["text"]

providers/kilo/models/~openai/gpt-latest.toml

@@ -16,10 +16,10 @@ type = "effort"
 values = ["none", "low", "medium", "high", "xhigh", "max"]
 
 [cost]
-input = 2
-output = 10
-cache_read = 0.2
-cache_write = 2.5
+input = 2.5
+output = 15
+cache_read = 0.25
+cache_write = 3.125
 
 [limit]
 context = 1_050_000

providers/llmgateway-providers/models/deepinfra/mimo-v2.5-pro.toml

@@ -1,25 +0,0 @@
-base_model = "xiaomi/mimo-v2.5-pro"
-name = "MiMo V2.5 Pro (DeepInfra)"
-tool_call = false
-structured_output = false
-
-[interleaved]
-field = "reasoning_content"
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "minimal", "low", "medium", "high", "xhigh", "max"]
-
-[cost]
-input = 1
-output = 3
-cache_read = 0.2
-
-[[cost.tiers]]
-tier = { type = "context", size = 256_000 }
-input = 2
-output = 6
-cache_read = 0.4
-
-[limit]
-output = 16_384

providers/llmgateway-providers/models/deepinfra/mimo-v2.5.toml

@@ -1,25 +0,0 @@
-base_model = "xiaomi/mimo-v2.5"
-name = "MiMo V2.5 (DeepInfra)"
-structured_output = true
-
-[interleaved]
-field = "reasoning_content"
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "minimal", "low", "medium", "high", "xhigh", "max"]
-
-[cost]
-input = 0.4
-output = 2
-cache_read = 0.08
-
-[[cost.tiers]]
-tier = { type = "context", size = 256_000 }
-input = 0.8
-output = 4
-cache_read = 0.16
-
-[limit]
-context = 262_144
-output = 16_384

providers/llmgateway-providers/models/novita/mimo-v2.5-pro.toml

@@ -1,21 +0,0 @@
-base_model = "xiaomi/mimo-v2.5-pro"
-name = "MiMo V2.5 Pro (NovitaAI)"
-structured_output = false
-
-[interleaved]
-field = "reasoning_content"
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "minimal", "low", "medium", "high"]
-
-[cost]
-input = 0.522
-output = 1.044
-cache_read = 0.0043
-
-[[cost.tiers]]
-tier = { type = "context", size = 256_000 }
-input = 2
-output = 6
-cache_read = 0.4

providers/llmgateway-providers/models/novita/mimo-v2.5.toml

@@ -1,21 +0,0 @@
-base_model = "xiaomi/mimo-v2.5"
-name = "MiMo V2.5 (NovitaAI)"
-structured_output = false
-
-[interleaved]
-field = "reasoning_content"
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "low", "medium", "high"]
-
-[cost]
-input = 0.168
-output = 0.336
-cache_read = 0.0034
-
-[[cost.tiers]]
-tier = { type = "context", size = 256_000 }
-input = 0.8
-output = 4
-cache_read = 0.16

providers/llmgateway-providers/models/openai/gpt-4o-mini-search-preview.toml

@@ -0,0 +1,23 @@
+name = "GPT-4o Mini Search Preview (OpenAI)"
+description = "Compact GPT model for low-latency assistance and high-volume workloads"
+family = "gpt"
+release_date = "2024-10-01"
+last_updated = "2024-10-01"
+attachment = true
+reasoning = false
+temperature = false
+tool_call = false
+structured_output = false
+open_weights = false
+
+[cost]
+input = 0.15
+output = 0.6
+
+[limit]
+context = 128_000
+output = 16_384
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

providers/llmgateway-providers/models/openai/gpt-4o-search-preview.toml

@@ -0,0 +1,23 @@
+name = "GPT-4o Search Preview (OpenAI)"
+description = "GPT model for general reasoning, writing, coding, and tool-assisted tasks"
+family = "gpt"
+release_date = "2024-10-01"
+last_updated = "2024-10-01"
+attachment = true
+reasoning = false
+temperature = false
+tool_call = false
+structured_output = false
+open_weights = false
+
+[cost]
+input = 2.5
+output = 10
+
+[limit]
+context = 128_000
+output = 16_384
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

providers/llmgateway-providers/models/ranoai/deepseek-v4-flash.toml

@@ -1,17 +0,0 @@
-base_model = "deepseek/deepseek-v4-flash"
-name = "DeepSeek V4 Flash (RanoAI)"
-
-[interleaved]
-field = "reasoning_content"
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "minimal", "low", "medium", "high", "xhigh", "max"]
-
-[cost]
-input = 0.14
-output = 0.28
-cache_read = 0.028
-
-[limit]
-output = 393_216

providers/llmgateway-providers/models/vertex-openai/grok-4-6.toml

@@ -1,11 +0,0 @@
-base_model = "xai/grok-4.6"
-name = "Grok 4.6 (Vertex AI (OpenAI-compatible))"
-
-[[reasoning_options]]
-type = "effort"
-values = ["minimal", "low", "medium", "high"]
-
-[cost]
-input = 2
-output = 6
-cache_read = 0.5

providers/llmgateway/models/gpt-4o-mini-search-preview.toml

@@ -0,0 +1,23 @@
+name = "GPT-4o Mini Search Preview"
+description = "Compact GPT model for low-latency assistance and high-volume workloads"
+family = "gpt"
+release_date = "2024-10-01"
+last_updated = "2024-10-01"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = false
+structured_output = false
+open_weights = false
+
+[cost]
+input = 0.15
+output = 0.60
+
+[limit]
+context = 128_000
+output = 16_384
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

providers/llmgateway/models/gpt-4o-search-preview.toml

@@ -0,0 +1,23 @@
+name = "GPT-4o Search Preview"
+description = "GPT model for general reasoning, writing, coding, and tool-assisted tasks"
+family = "gpt"
+release_date = "2024-10-01"
+last_updated = "2024-10-01"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = false
+structured_output = false
+open_weights = false
+
+[cost]
+input = 2.50
+output = 10.00
+
+[limit]
+context = 128_000
+output = 16_384
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

providers/llmgateway/models/grok-4-6.toml

@@ -2,7 +2,7 @@ base_model = "xai/grok-4.6"
 
 [[reasoning_options]]
 type = "effort"
-values = ["minimal", "low", "medium", "high", "xhigh"]
+values = ["low", "medium", "high", "xhigh"]
 
 [cost]
 input = 2

providers/llmgateway/models/mimo-v2.5-pro.toml

@@ -8,7 +8,7 @@ field = "reasoning_content"
 
 [[reasoning_options]]
 type = "effort"
-values = ["none", "minimal", "low", "medium", "high", "xhigh", "max"]
+values = ["none", "low", "medium", "high"]
 
 [cost]
 input = 0.435
@@ -20,3 +20,6 @@ tier = { type = "context", size = 256_000 }
 input = 2
 output = 6
 cache_read = 0.4
+
+[limit]
+context = 1_000_000

providers/llmgateway/models/mimo-v2.5.toml

@@ -8,7 +8,7 @@ field = "reasoning_content"
 
 [[reasoning_options]]
 type = "effort"
-values = ["none", "minimal", "low", "medium", "high", "xhigh", "max"]
+values = ["none", "low", "medium", "high"]
 
 [cost]
 input = 0.14
@@ -20,3 +20,6 @@ tier = { type = "context", size = 256_000 }
 input = 0.8
 output = 4
 cache_read = 0.16
+
+[limit]
+context = 1_000_000

providers/merge-gateway/models/deepseek/deepseek-v4-pro-0423.toml

@@ -1,10 +0,0 @@
-base_model = "deepseek/deepseek-v4-pro-0423"
-structured_output = false
-reasoning_options = []
-
-[cost]
-input = 1.65
-output = 3.3
-
-[limit]
-output = 393_216

providers/merge-gateway/models/openai/gpt-5.6-sol.toml

@@ -8,6 +8,6 @@ type = "effort"
 values = ["none", "low", "medium", "high", "xhigh", "max"]
 
 [cost]
-input = 4
-output = 24
+input = 5
+output = 30
 cache_read = 0.5

providers/merge-gateway/models/qwen/qwen3.5-122b-a10b.toml

@@ -1,7 +1,6 @@
 # Source: https://api-gateway.merge.dev/v1/models?model=qwen%2Fqwen3.5-122b-a10b (accessed 2026-07-21)
 base_model = "alibaba/qwen3.5-122b-a10b"
 name = "Qwen3.5 122B A10B"
-attachment = false
 structured_output = false
 
 [[reasoning_options]]
@@ -13,8 +12,8 @@ output = 0.917
 cache_read = 0.023
 
 [limit]
-context = 131_072
-output = 32_768
+context = 256_000
+output = 64_000
 
 [modalities]
-input = ["text"]
+input = ["text", "image"]

providers/merge-gateway/models/qwen/qwen3.5-35b-a3b.toml

@@ -2,6 +2,7 @@
 # Selected route reasoning.controls = ["thinking"], disable_supported = true.
 base_model = "alibaba/qwen3.5-35b-a3b"
 name = "Qwen3.5 35B A3B"
+attachment = false
 structured_output = false
 
 [[reasoning_options]]
@@ -13,8 +14,8 @@ output = 0.459
 cache_read = 0.020357
 
 [limit]
-context = 256_000
-output = 64_000
+context = 131_072
+output = 32_768
 
 [modalities]
-input = ["text", "image"]
+input = ["text"]

providers/nano-gpt/models/Gemma-4-31B-MeroMero-v2.toml

@@ -1,25 +0,0 @@
-name = "Gemma 4 31B MeroMero v2"
-description = "Gemma 4 31B MeroMero v2 is a LoRA finetune for emotive dialogue, relationship scenes, creative writing, and multimodal roleplay."
-family = "gemma"
-release_date = "2026-08-23"
-last_updated = "2026-08-23"
-attachment = true
-reasoning = true
-tool_call = true
-structured_output = false
-open_weights = true
-reasoning_options = []
-
-[cost]
-input = 0.08
-output = 0.33
-cache_read = 0.04
-
-[limit]
-context = 65_536
-input = 65_536
-output = 32_768
-
-[modalities]
-input = ["text", "image"]
-output = ["text"]

providers/nano-gpt/models/amazon/nova-micro-v1.toml

@@ -0,0 +1,24 @@
+name = "Amazon Nova Micro 1.0"
+description = "Efficient model for low-latency assistance, extraction, and routine automation"
+family = "nova-micro"
+release_date = "2024-01-01"
+last_updated = "2024-12-03"
+attachment = false
+reasoning = false
+tool_call = false
+structured_output = false
+open_weights = false
+
+[cost]
+input = 0.0357
+output = 0.1394
+cache_read = 0.01785
+
+[limit]
+context = 128_000
+input = 128_000
+output = 5_120
+
+[modalities]
+input = ["text"]
+output = ["text"]

providers/nano-gpt/models/deepseek/deepseek-v4-flash-vision-exp.toml

@@ -1,14 +0,0 @@
-base_model = "deepseek/deepseek-v4-flash-vision-exp"
-
-[[reasoning_options]]
-type = "effort"
-values = ["none", "low", "high", "max"]
-
-[cost]
-input = 0.22
-output = 0.66
-cache_read = 0.007
-
-[limit]
-context = 1_048_576
-input = 1_048_576

providers/nano-gpt/models/google/gemini-flash-latest.toml

@@ -1,16 +1,22 @@
 # Not included in subscription
 # Tracks Gemini 3.5 Flash (current flash-latest target).
 base_model = "google/gemini-flash-latest"
+tool_call = false
+structured_output = false
 
 [[reasoning_options]]
 type = "effort"
 values = ["minimal", "low", "medium", "high"]
 
 [cost]
-input = 0.375
-output = 1.875
-cache_read = 0.0375
-cache_write = 0.020833
+input = 0.75
+output = 3.75
+cache_read = 0.075
+cache_write = 0.041667
 
 [limit]
-input = 1_048_576
+context = 1_048_756
+input = 1_048_756
+
+[modalities]
+input = ["text", "image", "audio"]

providers/nano-gpt/models/google/gemma-4-26b-a4b-it.toml

@@ -6,9 +6,9 @@ name = "Gemma 4 26B A4B"
 type = "toggle"
 
 [cost]
-input = 0.08
-output = 0.33
-cache_read = 0.04
+input = 0.13
+output = 0.4
+cache_read = 0.065
 
 [limit]
 input = 262_144