[abacus] Add reasoning options

[wafer.ai] Add reasoning options

.github/workflows/close-stale-pull-requests.yml

@@ -0,0 +1,68 @@
+name: Close stale pull requests
+
+on:
+  schedule:
+    - cron: "17 3 * * *"
+  workflow_dispatch:
+
+permissions:
+  issues: write
+  pull-requests: write
+
+jobs:
+  close-stale-pull-requests:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/github-script@v8
+        env:
+          REVIEWER: rekram1-node
+        with:
+          script: |
+            const { owner, repo } = context.repo
+            const now = Date.now()
+            const weekAgo = now - 7 * 24 * 60 * 60 * 1000
+            const monthAgo = now - 30 * 24 * 60 * 60 * 1000
+
+            const pulls = await github.paginate(github.rest.pulls.list, {
+              owner,
+              repo,
+              state: "open",
+              per_page: 100,
+            })
+
+            const feedbackPulls = new Set()
+            for (const qualifier of ["commenter", "reviewed-by"]) {
+              const results = await github.paginate(
+                github.rest.search.issuesAndPullRequests,
+                {
+                  q: `repo:${owner}/${repo} is:pr is:open ${qualifier}:${process.env.REVIEWER}`,
+                  per_page: 100,
+                },
+              )
+
+              for (const result of results) feedbackPulls.add(result.number)
+            }
+
+            for (const pull of pulls) {
+              const updatedAt = Date.parse(pull.updated_at)
+              const monthStale = updatedAt < monthAgo
+              const feedbackStale = updatedAt < weekAgo && feedbackPulls.has(pull.number)
+              if (!monthStale && !feedbackStale) continue
+
+              const reason = monthStale
+                ? "it has not been updated in 30 days"
+                : `it has not been updated in 7 days after feedback from @${process.env.REVIEWER}`
+
+              await github.rest.issues.createComment({
+                owner,
+                repo,
+                issue_number: pull.number,
+                body: `Closing this pull request as stale because ${reason}. Feel free to reopen it or submit a new pull request if the work is resumed.`,
+              })
+              await github.rest.pulls.update({
+                owner,
+                repo,
+                pull_number: pull.number,
+                state: "closed",
+              })
+            }

.github/workflows/sync-models.yml

@@ -63,7 +63,9 @@ jobs:
       - name: Sync model catalogs
         run: bun models:sync ${{ matrix.provider }}
         env:
+          BASETEN_API_KEY: ${{ secrets.BASETEN_API_KEY }}
           OPENROUTER_API_KEY: ${{ secrets.OPENROUTER_API_KEY }}
+          VENICE_API_KEY: ${{ secrets.VENICE_API_KEY }}
           GOOGLE_API_KEY: ${{ secrets.GOOGLE_API_KEY }}
           GEMINI_API_KEY: ${{ secrets.GEMINI_API_KEY }}
           GOOGLE_GENERATIVE_AI_API_KEY: ${{ secrets.GOOGLE_GENERATIVE_AI_API_KEY }}
@@ -81,7 +83,7 @@ jobs:
           LABELS: automation,model-sync,provider:${{ matrix.provider }}
           TITLE: "chore(sync): update ${{ matrix.name }} model catalog"
         run: |
-          if [ -z "$(git status --porcelain -- providers)" ]; then
+          if [ -z "$(git status --porcelain -- models providers)" ]; then
             echo "No model catalog changes found."
             exit 0
           fi
@@ -90,7 +92,7 @@ jobs:
           git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
           git fetch --no-tags --depth=1 origin "+refs/heads/$BRANCH:refs/remotes/origin/$BRANCH" || true
           git checkout -B "$BRANCH"
-          git add providers
+          git add models providers
           git commit -m "$TITLE"
           git push --force-with-lease origin "$BRANCH"
 

README.md

@@ -220,6 +220,8 @@ Rules:
 
 Use `base_model` when the wrapper model is materially the same as the source model and only differs by provider-specific pricing, limits, modalities, provider request shape, or lifecycle flags.
 
+Sync and generator scripts should preserve existing `base_model` / `base_model_omit` fields when updating provider TOMLs. Do not use legacy `[extends]` tables.
+
 #### 4. Submit a Pull Request
 
 1. Fork this repo

labs/alibaba/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M37.9998 23.021C33.7998 25.2889 29.5698 27.3649 24.8614 28.3069C23.8114 28.5154 22.6474 28.5154 21.5809 28.3714C20.5639 28.2439 20.0554 27.3484 20.4169 26.4064C20.7619 25.5289 21.2209 24.635 21.8119 23.9C23.0899 22.3025 24.5329 20.849 25.8289 19.268C26.6203 18.2991 27.3335 17.2689 27.9618 16.187C28.4208 15.4205 28.2078 14.4935 27.4038 14.111C26.0584 13.4556 24.6154 12.9936 23.1889 12.4986C23.0239 12.4341 22.7779 12.6096 22.4509 12.7221C22.8604 13.0881 23.1559 13.3596 23.5654 13.727C19.3339 14.447 15.3305 15.467 11.4455 16.874C11.4275 16.9535 11.396 17.0165 11.411 17.0495C11.9855 17.927 11.723 18.5975 10.886 19.1405C10.5611 19.3531 10.2732 19.6176 10.034 19.9235C12.593 20.6735 14.873 20.243 17.0539 18.821C16.9234 18.6305 16.7914 18.455 16.6609 18.263C17.4799 18.407 17.9719 18.854 18.0379 19.556C18.0544 19.7165 17.9569 19.8755 17.9074 20.036C17.7919 19.907 17.6449 19.781 17.5474 19.6355C17.4799 19.5395 17.4634 19.4285 17.4154 19.268C14.8235 20.993 12.035 21.425 8.96751 20.531C8.96751 21.137 8.93451 21.6485 8.98401 22.1435C9.01701 22.574 8.83701 22.766 8.44401 22.9895C7.55752 23.5325 6.63803 24.092 5.90003 24.8105C5.01504 25.6879 5.34354 26.7589 6.54053 27.2059C7.90102 27.7159 9.329 27.7309 10.7555 27.5569C12.4445 27.3484 14.1005 27.0769 15.9394 26.8219C13.79 27.8269 11.6735 28.5319 9.4445 28.8169C7.88452 29.0269 6.32753 29.1379 4.78554 28.6909C2.57156 28.0684 1.58607 26.4394 2.16057 24.251C2.70206 22.2065 4.01455 20.5775 5.42454 19.076C10.133 14.078 16.0864 11.5401 22.9744 11.0286C24.5824 10.9176 26.2069 11.1246 27.7143 11.7951C29.8308 12.7536 30.7173 14.78 29.6838 16.826C29.0118 18.1835 28.0758 19.4285 27.1413 20.6585C26.2234 21.872 25.1899 22.9895 24.2224 24.155C23.9434 24.506 23.6809 24.875 23.4679 25.2724C23.0569 26.0224 23.3359 26.5174 24.2059 26.4394C26.0254 26.2624 27.8808 26.1199 29.6358 25.6729C32.2098 25.0174 34.7193 24.092 37.2618 23.2775C37.5243 23.213 37.7703 23.117 37.9998 23.0225V23.021Z" fill="currentColor"/>
+</svg>

labs/anthropic/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M26.9568 9.88184H22.1265L30.7753 31.7848H35.4917L26.9568 9.88184ZM13.028 9.88184L4.4917 31.7848H9.32203L11.2305 27.1793H20.2166L22.0126 31.6724H26.8444L18.0832 9.88184H13.028ZM12.5783 23.1361L15.4987 15.3853L18.5315 23.1361H12.5783Z" fill="currentColor"/>
+</svg>

labs/cohere/logo.svg

@@ -0,0 +1,5 @@
+<svg width="24" height="24" viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path d="M9.14882 13.5552C9.58082 13.5552 10.4448 13.5336 11.6544 13.0368C13.0584 12.4536 15.8232 11.4168 17.832 10.3368C19.236 9.5808 19.8408 8.5872 19.8408 7.248C19.8408 5.412 18.3504 3.9 16.4928 3.9H8.71682C6.06002 3.9 3.90002 6.06 3.90002 8.7168C3.90002 11.3736 5.93042 13.5552 9.14882 13.5552Z" fill="currentColor"/>
+<path d="M10.4664 16.86C10.4664 15.564 11.244 14.376 12.4536 13.8792L14.8944 12.864C17.3784 11.8488 20.1 13.6632 20.1 16.3416C20.1 18.4152 18.4152 20.1 16.3416 20.1H13.6848C11.9136 20.1 10.4664 18.6528 10.4664 16.86Z" fill="currentColor"/>
+<path d="M6.68642 14.1816C5.15282 14.1816 3.90002 15.4344 3.90002 16.968V17.3352C3.90002 18.8472 5.15282 20.1 6.68642 20.1C8.22003 20.1 9.47283 18.8472 9.47283 17.3136V16.9464C9.45123 15.4344 8.22003 14.1816 6.68642 14.1816Z" fill="currentColor"/>
+</svg>

labs/deepseek/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M35.6638 9.91965C35.3251 9.75432 35.1785 10.0703 34.9811 10.2316C34.9131 10.2836 34.8558 10.3516 34.7985 10.413C34.3025 10.9423 33.7238 11.289 32.9678 11.2476C31.8625 11.1863 30.9186 11.533 30.0839 12.3783C29.9066 11.3356 29.3173 10.7143 28.4213 10.3143C27.9519 10.1063 27.4773 9.89965 27.148 9.44766C26.9186 9.12633 26.856 8.76767 26.7413 8.41568C26.668 8.20235 26.5946 7.98502 26.3506 7.94902C26.084 7.90769 25.98 8.13035 25.876 8.31702C25.4587 9.07967 25.2973 9.91965 25.3133 10.7703C25.3493 12.6849 26.1573 14.2102 27.764 15.2942C27.9466 15.4182 27.9933 15.5435 27.9359 15.7249C27.8266 16.0982 27.696 16.4609 27.5813 16.8355C27.508 17.0742 27.3986 17.1248 27.1426 17.0222C26.2777 16.6504 25.4919 16.1164 24.828 15.4489C23.6854 14.3449 22.6534 13.1263 21.3654 12.1716C21.067 11.9511 20.7606 11.7416 20.4468 11.5436C19.1335 10.2676 20.6201 9.21967 20.9641 9.09567C21.3241 8.965 21.0881 8.51968 19.9254 8.52501C18.7628 8.53035 17.6988 8.91834 16.3428 9.43699C16.1413 9.51421 15.934 9.57529 15.7229 9.61966C14.4557 9.38091 13.1598 9.33506 11.8789 9.48366C9.36565 9.76365 7.35902 10.953 5.88305 12.9809C4.10975 15.4182 3.69243 18.1888 4.20308 21.0768C4.74041 24.122 6.29504 26.6433 8.683 28.6139C11.1603 30.6579 14.0122 31.6592 17.2668 31.4672C19.2428 31.3539 21.4441 31.0886 23.9254 28.9873C24.552 29.2993 25.208 29.4233 26.2986 29.5166C27.1386 29.5953 27.9466 29.4766 28.5719 29.3459C29.5519 29.1379 29.4839 28.23 29.1306 28.0646C26.2573 26.726 26.888 27.2713 26.3133 26.83C27.7746 25.102 29.9746 23.3074 30.8359 17.4928C30.9026 17.0302 30.8452 16.7395 30.8359 16.3662C30.8306 16.1395 30.8826 16.0502 31.1426 16.0249C31.8639 15.95 32.5637 15.7349 33.2025 15.3915C35.0638 14.3742 35.8158 12.7049 35.9931 10.7023C36.0198 10.3956 35.9878 10.081 35.6638 9.91965ZM19.4414 27.9433C16.6562 25.754 15.3055 25.0327 14.7482 25.0634C14.2256 25.0954 14.3202 25.6913 14.4349 26.0807C14.5549 26.4647 14.7109 26.7286 14.9295 27.066C15.0815 27.2886 15.1855 27.6206 14.7789 27.87C13.8816 28.4246 12.3229 27.6833 12.2496 27.6473C10.435 26.578 8.91632 25.1673 7.84834 23.2381C6.81637 21.3808 6.21638 19.3888 6.11771 17.2622C6.09105 16.7475 6.24171 16.5662 6.7537 16.4729C7.42583 16.3442 8.11451 16.3267 8.79233 16.4209C11.6349 16.8368 14.0536 18.1075 16.0828 20.1194C17.2402 21.2661 18.1161 22.6354 19.0188 23.974C19.9788 25.3953 21.0108 26.75 22.3254 27.8593C22.7894 28.2486 23.1587 28.5446 23.5134 28.7619C22.4441 28.8819 20.6601 28.9086 19.4414 27.9433ZM20.7748 19.3568C20.7745 19.2906 20.7904 19.2253 20.8211 19.1666C20.8517 19.1078 20.8962 19.0575 20.9507 19.0198C21.0052 18.9821 21.068 18.9583 21.1337 18.9503C21.1995 18.9424 21.2662 18.9505 21.3281 18.9741C21.407 19.0024 21.475 19.0546 21.5228 19.1235C21.5706 19.1923 21.5958 19.2743 21.5947 19.3581C21.5949 19.4123 21.5843 19.4659 21.5636 19.5159C21.5428 19.5659 21.5123 19.6113 21.4738 19.6494C21.4354 19.6875 21.3897 19.7176 21.3395 19.7378C21.2893 19.7581 21.2356 19.7682 21.1814 19.7675C21.1277 19.7676 21.0745 19.7571 21.0248 19.7365C20.9752 19.7158 20.9302 19.6855 20.8925 19.6473C20.8548 19.609 20.825 19.5636 20.805 19.5138C20.785 19.4639 20.7739 19.4105 20.7748 19.3568ZM24.9213 21.4848C24.6547 21.5928 24.3893 21.6861 24.1347 21.6981C23.7516 21.7114 23.3756 21.5918 23.0707 21.3594C22.7054 21.0528 22.4441 20.8821 22.3347 20.3488C22.297 20.0881 22.3042 19.823 22.3561 19.5648C22.4494 19.1288 22.3454 18.8488 22.0374 18.5955C21.7881 18.3875 21.4694 18.3302 21.1201 18.3302C21.0005 18.3232 20.8843 18.2875 20.7814 18.2262C20.6348 18.1542 20.5148 17.9728 20.6294 17.7488C20.6668 17.6768 20.8428 17.5008 20.8854 17.4688C21.3601 17.1995 21.9081 17.2875 22.4134 17.4902C22.8827 17.6822 23.2374 18.0342 23.748 18.5328C24.2694 19.1341 24.364 19.3008 24.6613 19.7515C24.896 20.1048 25.1093 20.4674 25.2547 20.8821C25.344 21.1421 25.2293 21.3541 24.9213 21.4848Z" fill="currentColor"/>
+</svg>

labs/google/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M37 20.034C27.8809 20.5837 20.5808 27.8809 20.0326 37H19.966C19.4163 27.8809 12.1177 20.5837 3 20.034V19.9674C12.1191 19.4163 19.4163 12.1191 19.966 3H20.0326C20.5822 12.1191 27.8809 19.4163 37 19.9674V20.034Z" fill="currentColor"/>
+</svg>

labs/meta/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M27.1942 9.03509C24.4881 9.03509 22.3726 11.0731 20.4574 13.6623C17.8258 10.3114 15.6255 9.03509 12.9925 9.03509C7.62404 9.03509 3.51001 16.0234 3.51001 23.4181C3.51001 28.0453 5.74831 30.9649 9.49831 30.9649C12.1971 30.9649 14.1387 29.693 17.5904 23.6594C17.5904 23.6594 19.029 21.1199 20.0173 19.3699C20.3643 19.9293 20.7298 20.5327 21.1138 21.1798L22.7322 23.902C25.8843 29.1769 27.6416 30.9649 30.8229 30.9649C34.4778 30.9649 36.51 28.0058 36.51 23.2822C36.51 15.538 32.3039 9.03509 27.1942 9.03509ZM14.9574 22.0263C12.1606 26.4123 11.1928 27.3962 9.63574 27.3962C8.03194 27.3962 7.07872 25.9883 7.07872 23.4781C7.07872 18.1096 9.75562 12.6199 12.9471 12.6199C14.6752 12.6199 16.1197 13.617 18.3316 16.7836C16.2308 20.0058 14.9574 22.0263 14.9574 22.0263ZM25.5202 21.4751L23.5831 18.2456C23.0969 17.4514 22.5938 16.6676 22.0743 15.8947C23.8185 13.2032 25.2556 11.8611 26.9676 11.8611C30.5202 11.8611 33.3638 17.095 33.3638 23.5219C33.3638 25.9722 32.5612 27.3947 30.8989 27.3947C29.3053 27.3947 28.5451 26.3421 25.5188 21.4737" fill="currentColor"/>
+</svg>

labs/minimax/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M17.8758 9.20865C17.8758 8.59461 17.3777 8.09634 16.7663 8.09634C16.155 8.09634 15.6567 8.59575 15.6567 9.20865V27.6446C15.6567 29.0714 14.4985 30.2324 13.0755 30.2324C11.6523 30.2324 10.4941 29.0714 10.4941 27.6446V15.8167C10.4941 15.2027 9.99591 14.7044 9.38453 14.7044C8.77316 14.7044 8.275 15.2038 8.275 15.8167V20.8301C8.275 22.2567 7.11678 23.4179 5.69364 23.4179C4.2705 23.4179 3.1123 22.2567 3.1123 20.8301V19.0129C3.1123 18.6054 3.44177 18.2752 3.84822 18.2752C4.25467 18.2752 4.58413 18.6054 4.58413 19.0129V20.8301C4.58413 21.4441 5.08227 21.9424 5.69364 21.9424C6.30502 21.9424 6.80317 21.443 6.80317 20.8301V15.8167C6.80317 14.39 7.96139 13.2289 9.38453 13.2289C10.8077 13.2289 11.9659 14.39 11.9659 15.8167V27.6446C11.9659 28.2587 12.4641 28.7569 13.0755 28.7569C13.6868 28.7569 14.1849 28.2575 14.1849 27.6446V20.4123V9.20865C14.1849 7.78194 15.3431 6.62082 16.7663 6.62082C18.1894 6.62082 19.3476 7.78194 19.3476 9.20865V24.4746C19.3476 24.8821 19.0182 25.2123 18.6117 25.2123C18.2053 25.2123 17.8758 24.8821 17.8758 24.4746V9.20865ZM31.531 13.2289C30.1079 13.2289 28.9496 14.39 28.9496 15.8167V25.6969C28.9496 26.311 28.4515 26.8093 27.8401 26.8093C27.2287 26.8093 26.7306 26.3099 26.7306 25.6969V9.20865C26.7306 7.78194 25.5723 6.62082 24.1492 6.62082C22.7261 6.62082 21.5679 7.78194 21.5679 9.20865V30.1383C21.5679 30.7523 21.0697 31.2506 20.4583 31.2506C19.8469 31.2506 19.3488 30.7511 19.3488 30.1383V27.5471C19.3488 27.1396 19.0194 26.8093 18.6129 26.8093C18.2065 26.8093 17.877 27.1396 17.877 27.5471V30.1383C17.877 31.565 19.0352 32.7261 20.4583 32.7261C21.8815 32.7261 23.0397 31.565 23.0397 30.1383V9.20865C23.0397 8.59461 23.5378 8.09634 24.1492 8.09634C24.7605 8.09634 25.2587 8.59575 25.2587 9.20865V25.6969C25.2587 27.1237 26.417 28.2848 27.8401 28.2848C29.2632 28.2848 30.4215 27.1237 30.4215 25.6969V15.8167C30.4215 15.2027 30.9196 14.7044 31.531 14.7044C32.1424 14.7044 32.6405 15.2038 32.6405 15.8167V24.4746C32.6405 24.8821 32.97 25.2123 33.3764 25.2123C33.7829 25.2123 34.1123 24.8821 34.1123 24.4746V15.8167C34.1123 14.39 32.9541 13.2289 31.531 13.2289Z" fill="currentColor"/>
+</svg>

labs/mistral/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M8.92783 8.88101H13.357V13.3088H17.7861V17.738H17.7835H22.2152V13.3088H26.6418V8.88101H31.0722V26.5949H35.5V31.0241H22.2139V26.5962H17.7861V22.1671H13.3557V26.5949L17.7861 26.5962V31.0241H4.5V26.5949H8.92783V8.88101ZM22.2139 26.5962H26.6418V22.1671H22.2152V26.5962H22.2139Z" fill="currentColor"/>
+</svg>

labs/nvidia/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M16.1801 15.8791V14.0283C16.3472 14.0155 16.5271 14.0026 16.7199 14.0026C21.7966 13.8356 25.1254 18.3596 25.1254 18.3596C25.1254 18.3596 21.5267 23.3463 17.671 23.3463C17.1569 23.3463 16.6557 23.2692 16.1801 23.1021V17.4728C18.1465 17.7298 18.545 18.591 19.7274 20.5702L22.375 18.3468C22.375 18.3468 20.4343 15.8277 17.1955 15.8277C16.8356 15.8277 16.5014 15.8405 16.1801 15.8791ZM16.1801 9.75492V12.5246C16.3472 12.5118 16.5271 12.4989 16.7199 12.4861C23.763 12.2547 28.377 18.2696 28.377 18.2696C28.377 18.2696 23.0819 24.683 17.5939 24.683C17.0798 24.683 16.6171 24.6444 16.1801 24.5545V26.2767C16.5528 26.3152 16.9384 26.341 17.3625 26.341C22.4649 26.341 26.1664 23.7448 29.7523 20.6473C30.3435 21.1229 32.7597 22.2796 33.261 22.7937C29.8679 25.6341 21.9251 27.9346 17.4396 27.9346C17.0155 27.9346 16.5914 27.9089 16.1801 27.8704V30.2866H35.6001V9.75492H16.1801ZM16.1801 23.1021V24.5545C11.4376 23.7191 10.1266 18.7966 10.1266 18.7966C10.1266 18.7966 12.4015 16.2775 16.1801 15.8791V17.4728H16.1673C14.188 17.2414 12.6329 19.0922 12.6329 19.0922C12.6329 19.0922 13.5068 22.2025 16.1801 23.1021ZM7.77464 18.591C7.77464 18.591 10.5765 14.4525 16.1801 14.0283V12.5246C9.9724 13.0259 4.6001 18.2696 4.6001 18.2696C4.6001 18.2696 7.64612 27.0735 16.1801 27.8575V26.2767C9.90814 25.4798 7.77464 18.591 7.77464 18.591Z" fill="currentColor"/>
+</svg>

labs/openai/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M32.8377 17.282C33.2127 16.25 33.3072 15.218 33.2127 14.1875C33.1197 13.1571 32.7447 12.1251 32.2752 11.1876C31.4322 9.78209 30.2127 8.6571 28.8072 8.0001C27.3072 7.34461 25.7127 7.15711 24.1197 7.53211C23.3698 6.78212 22.5253 6.12512 21.5878 5.65713C20.6503 5.18913 19.5253 5.00013 18.4948 5.00013C16.8851 4.99074 15.3125 5.48246 13.9948 6.40712C12.6824 7.34311 11.7449 8.6571 11.2754 10.1571C10.1504 10.4376 9.21289 10.9071 8.27539 11.4696C7.4324 12.1251 6.77541 12.9696 6.21291 13.8126C5.36992 15.2195 5.08792 16.8125 5.27542 18.407C5.46399 19.9968 6.11605 21.496 7.1504 22.718C6.79608 23.7086 6.66795 24.7659 6.77541 25.8124C6.86991 26.8444 7.2449 27.8749 7.7129 28.8124C8.55739 30.2194 9.77538 31.3444 11.1824 31.9999C12.6824 32.6569 14.2753 32.8444 15.8698 32.4694C16.6198 33.2194 17.4628 33.8749 18.4003 34.3444C19.3378 34.8139 20.4628 34.9999 21.4948 34.9999C23.1043 35.0097 24.6769 34.5185 25.9947 33.5944C27.3072 32.6569 28.2447 31.3444 28.7127 29.8444C29.7719 29.6432 30.7682 29.1934 31.6197 28.5319C32.4627 27.8749 33.2127 27.1249 33.6822 26.1874C34.5251 24.7819 34.8071 23.1875 34.6196 21.5945C34.4322 20 33.8697 18.5015 32.8377 17.282ZM21.5878 33.0304C20.0878 33.0304 18.9628 32.5609 17.9323 31.7179C17.9323 31.7179 18.0253 31.6234 18.1198 31.6234L24.1197 28.1554C24.2862 28.0803 24.4196 27.9469 24.4947 27.7804C24.5698 27.636 24.6021 27.4731 24.5877 27.3109V18.875L27.1197 20.375V27.3124C27.1455 28.0547 27.0215 28.7945 26.755 29.4878C26.4885 30.181 26.085 30.8134 25.5687 31.3473C25.0523 31.8811 24.4337 32.3054 23.7497 32.5949C23.0658 32.8843 22.3305 33.0314 21.5878 33.0304ZM9.49488 27.8749C8.83789 26.7499 8.55739 25.4374 8.83789 24.125C8.83789 24.125 8.93239 24.2195 9.02539 24.2195L15.0253 27.6874C15.1693 27.7638 15.3325 27.7966 15.4948 27.7819C15.6823 27.7819 15.8698 27.7819 15.9628 27.6874L23.2753 23.4695V26.3749L17.1823 29.9374C16.5506 30.3042 15.8527 30.5427 15.1287 30.6393C14.4046 30.7358 13.6686 30.6884 12.9629 30.4999C11.4629 30.1249 10.2449 29.1874 9.49488 27.8749ZM7.9004 14.8445C8.56239 13.7234 9.58826 12.8627 10.8074 12.4056V19.532C10.8074 19.718 10.8074 19.907 10.9004 20C10.9755 20.1665 11.1089 20.2998 11.2754 20.375L18.5878 24.5944L16.0573 26.0944L10.0574 22.625C9.41842 22.2639 8.85742 21.7797 8.40684 21.2004C7.95627 20.6211 7.62506 19.9582 7.4324 19.25C7.05741 17.8445 7.1504 16.157 7.9004 14.8445ZM28.6197 19.625L21.3073 15.407L23.8377 13.9071L29.8377 17.375C30.7752 17.9375 31.5252 18.6875 31.9947 19.625C32.4642 20.5625 32.7447 21.5945 32.6502 22.7195C32.5603 23.7755 32.1699 24.7837 31.5252 25.6249C30.8697 26.4694 30.0252 27.1249 28.9947 27.4999V20.375C28.9947 20.1875 28.9947 20 28.9002 19.907C28.9002 19.907 28.8072 19.718 28.6197 19.625ZM31.1502 15.875C31.1502 15.875 31.0572 15.782 30.9627 15.782L24.9627 12.3126C24.7752 12.2196 24.6822 12.2196 24.4947 12.2196C24.3072 12.2196 24.1197 12.2196 24.0252 12.3126L16.7128 16.532V13.6251L22.8073 10.0626C23.7448 9.50009 24.7752 9.31259 25.9002 9.31259C26.9322 9.31259 27.9627 9.68759 28.9002 10.3446C29.7447 11.0001 30.4947 11.8446 30.8697 12.7821C31.2447 13.7196 31.3377 14.8445 31.1502 15.875ZM15.4003 21.125L12.8699 19.625V12.5946C12.8699 11.5626 13.1503 10.4376 13.7128 9.59459C14.2753 8.6571 15.1198 8.0001 16.0573 7.53211C17.0127 7.05249 18.0956 6.88812 19.1503 7.06261C20.1823 7.15711 21.2128 7.62511 22.0573 8.2821C22.0573 8.2821 21.9628 8.3751 21.8698 8.3751L15.8698 11.8446C15.7033 11.9197 15.57 12.0531 15.4948 12.2196C15.4003 12.4071 15.4003 12.5001 15.4003 12.6876V21.125ZM16.7128 18.125L19.9948 16.25L23.2753 18.125V21.875L19.9948 23.75L16.7128 21.875V18.125Z" fill="currentColor"/>
+</svg>

labs/perplexity/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path d="M17.2642 2.8689L12.042 8.0961M17.2642 2.8689V8.0961H12.042M17.2642 2.8689V4.30027M12.042 8.0961L6.81809 2.8689V8.0961H12.042ZM12.042 8.0961L17.2642 13.3225V20.8159L12.042 15.5887M12.042 8.0961V15.5887M12.042 8.0961L6.81892 13.3225M12.0296 2.1V21.9M12.042 15.5887L6.81892 20.8159V13.3225M6.81892 13.3225L6.81809 15.559H4.57739V8.09527H12.0412L6.81892 13.3225ZM11.9859 8.09527L17.2081 13.3225V15.559H19.4497V8.09527H11.9859Z" stroke="currentColor" stroke-width="0.825" stroke-miterlimit="10"/>
+</svg>

labs/tencent/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 32 32" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path fill-rule="evenodd" clip-rule="evenodd" d="M18.0835 25.1541C19.5821 26.7684 21.1466 28.1292 22.6694 29.2117L21.5103 30.8416C19.8992 29.6963 18.2543 28.2675 16.6802 26.5808C17.1455 26.1486 17.6143 25.6722 18.0835 25.1541ZM9.99365 2.74979C11.7639 1.73255 13.8054 2.24897 15.6479 3.32304C17.52 4.41448 19.4573 6.22477 21.2603 8.33866L19.7378 9.63651C18.0045 7.60434 16.238 5.98161 14.6411 5.05058C13.0147 4.10247 11.8203 4.00682 10.9897 4.48417C10.0366 5.03255 9.41678 6.47496 9.78271 9.17167C9.86549 9.78127 9.99965 10.4345 10.1812 11.1277C11.6144 10.9497 13.0985 10.8513 14.6167 10.8543C17.7104 10.8604 20.4741 11.1506 22.7944 11.6805L22.7964 11.6814C23.2 11.7609 23.5914 11.858 23.9702 11.9656C23.9685 11.9703 23.9661 11.9746 23.9644 11.9793C24.661 12.1766 25.3101 12.3961 25.9067 12.6404C28.6066 13.7462 30.6077 15.4947 30.603 17.8855C30.599 19.9274 29.1314 21.4372 27.2798 22.4959C25.3986 23.5714 22.8625 24.344 20.1304 24.8484L19.7671 22.8807C22.3935 22.3957 24.6819 21.677 26.2866 20.7596C27.9209 19.8251 28.6011 18.8396 28.603 17.8816C28.6051 16.7818 27.6669 15.5237 25.1479 14.492C24.5672 14.2542 23.9208 14.0387 23.2134 13.8465C22.6517 15.1437 22.0035 16.4444 21.2612 17.7244C18.9392 21.7281 16.437 24.776 14.0688 26.6082C11.7614 28.3932 9.24631 29.2529 7.17822 28.0535C5.41209 27.0291 4.83828 25.0026 4.84717 22.8699C4.85632 20.7032 5.45524 18.1208 6.38428 15.5027L8.27002 16.1717C7.37685 18.6887 6.85497 21.0304 6.84717 22.8787C6.83939 24.7607 7.35367 25.8424 8.18213 26.323C9.13349 26.8747 10.6925 26.6915 12.8452 25.0262C13.3191 24.6595 13.8045 24.2295 14.3003 23.742C13.4493 22.6129 12.6395 21.4021 11.897 20.1101C9.59084 16.0976 8.20303 12.4071 7.80029 9.44022C7.40803 6.54936 7.92116 3.94134 9.99365 2.74979ZM14.6128 12.8543C13.3044 12.8518 12.0235 12.9291 10.7827 13.0701C11.4353 14.9013 12.3808 16.9381 13.6313 19.1141C14.2698 20.2249 14.9606 21.2715 15.6841 22.2537C16.9553 20.7654 18.2592 18.9138 19.5308 16.7215C20.1679 15.623 20.7315 14.5102 21.2271 13.4012C19.3246 13.0558 17.1033 12.8592 14.6128 12.8543ZM6.23193 11.8416C6.39799 12.459 6.59817 13.0982 6.83154 13.7566C4.8717 14.2207 3.07218 14.836 1.49854 15.5398L0.682129 13.7137C2.33167 12.9759 4.20224 12.3309 6.23193 11.8416ZM24.8901 3.62675L25.8853 3.70976C25.7206 5.68459 25.2981 7.84221 24.6128 10.0701C24.1572 9.9381 23.6801 9.8164 23.1831 9.71854C23.0101 9.6845 22.8337 9.6559 22.6567 9.62577C23.3278 7.47916 23.7373 5.41176 23.8931 3.54276L24.8901 3.62675Z" fill="currentColor"/>
+</svg>

labs/xai/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M12.4579 15.6036L26.1529 35H20.0656L6.37059 15.6036H12.4579ZM12.4524 26.3764L15.4974 30.6909L12.4551 35H6.36377L12.4524 26.3764ZM33.6365 7.15727V35H28.647V14.2236L33.6365 7.15727ZM33.6365 5L20.0656 24.2205L17.0206 19.9073L27.5451 5H33.6365Z" fill="currentColor"/>
+</svg>

labs/xiaomi/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M17.3598 5.03522C13.6378 5.20577 11.3905 5.78765 9.47432 7.07179C8.10992 7.97471 7.10668 9.14849 6.38435 10.6935C5.59179 12.389 5.24066 14.255 5.09017 17.7262C4.90959 21.6087 5.17043 25.4812 5.75231 27.5378C6.27399 29.4039 7.10668 30.8285 8.38079 32.0223C9.95588 33.487 11.7216 34.2595 14.4604 34.6709C16.9685 35.0421 22.7773 35.0621 25.3154 34.711C28.4054 34.2896 30.3718 33.4168 31.987 31.7414C32.7896 30.9187 33.1608 30.377 33.6824 29.2835C34.4951 27.5679 34.8261 25.7922 34.9867 22.2407C35.107 19.7026 34.9867 16.0909 34.7358 14.4055C34.1439 10.4527 32.4585 7.88441 29.519 6.39962C27.964 5.62713 25.6766 5.17567 22.4963 5.02519C20.0986 4.91483 19.7776 4.91483 17.3598 5.03522ZM21.0617 14.3854C22.386 14.6964 23.0281 15.1378 23.4494 16.0407C23.8808 16.9637 23.951 17.7262 23.951 21.8896V25.8022L22.5264 25.7721L21.0918 25.742L21.0417 21.9297C20.9915 17.7663 20.9714 17.6359 20.3895 17.1844C19.8679 16.7731 19.4264 16.7229 16.5772 16.7129H13.8685L13.8183 21.2275L13.7682 25.742H10.9591L10.929 20.0336C10.909 15.5291 10.939 14.2951 11.0293 14.2349C11.0996 14.1847 13.2365 14.1647 15.7747 14.1847C19.4967 14.2249 20.52 14.265 21.0617 14.3854ZM28.9472 14.3252C29.0174 14.4657 29.0475 25.3708 28.9773 25.722C28.9773 25.7621 28.3252 25.7822 27.5426 25.7721L26.108 25.742L26.0779 20.0838C26.0679 15.9906 26.0879 14.3854 26.1682 14.2951C26.2585 14.1847 26.6096 14.1546 27.5727 14.1546C28.6964 14.1546 28.8669 14.1747 28.9472 14.3252ZM18.9349 22.2307V25.8022L17.4601 25.7721L15.9753 25.742L15.9452 22.331C15.9352 20.455 15.9452 18.8598 15.9753 18.7996C16.0054 18.6993 16.3967 18.6692 17.4802 18.6692H18.9349V22.2307Z" fill="currentColor"/>
+</svg>

labs/zhipuai/logo.svg

@@ -0,0 +1,3 @@
+<svg width="24" height="24" viewBox="0 0 40 40" xmlns="http://www.w3.org/2000/svg">
+<path d="M20.1312 7.50002L17.4088 11.1913H5.81625L8.5375 7.50002H20.1325H20.1312ZM34.0675 28.81L31.3475 32.5H19.795L22.5125 28.81H34.0675ZM35 7.50002L16.58 32.5H5L23.42 7.50002H35Z" fill="currentColor"/>
+</svg>

models/alibaba/qwen-max.toml

@@ -16,3 +16,10 @@ output = 8_192
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 21.8
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-01-28"

models/alibaba/qwen2-5-vl-72b-instruct.toml

@@ -16,3 +16,7 @@ output = 8_192
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen2.5-VL-72B-Instruct"

models/alibaba/qwen3-235b-a22b.toml

@@ -16,3 +16,21 @@ output = 16_384
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-235B-A22B"
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 59.6
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-05-09"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 21.41
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/alibaba/qwen3-32b.toml

@@ -16,3 +16,14 @@ output = 16_384
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-32B"
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 40.0
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-05-08"

models/alibaba/qwen3-coder-30b-a3b-instruct.toml

@@ -16,3 +16,28 @@ output = 65_536
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 19.4
+metric = "index"
+source = "https://openrouter.ai/qwen/qwen3-coder-30b-a3b-instruct/benchmarks"
+date = "2026-06-02"
+
+[[benchmarks]]
+name = "SciCode"
+score = 27.8
+metric = "percent correct"
+source = "https://openrouter.ai/qwen/qwen3-coder-30b-a3b-instruct/benchmarks"
+date = "2026-06-02"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 15.2
+metric = "success rate"
+source = "https://openrouter.ai/qwen/qwen3-coder-30b-a3b-instruct/benchmarks"
+date = "2026-06-02"

models/alibaba/qwen3-coder-480b-a35b-instruct.toml

@@ -16,3 +16,14 @@ output = 65_536
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 38.7
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/alibaba/qwen3-coder-plus.toml

@@ -7,7 +7,7 @@ reasoning = false
 temperature = true
 tool_call = true
 knowledge = "2025-04"
-open_weights = true
+open_weights = false
 
 [limit]
 context = 1_048_576

models/alibaba/qwen3-max.toml

@@ -16,3 +16,24 @@ output = 65_536
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 26.4
+metric = "index"
+source = "https://openrouter.ai/qwen/qwen3-max/benchmarks"
+date = "2026-05-30"
+
+[[benchmarks]]
+name = "SciCode"
+score = 38.3
+metric = "percent correct"
+source = "https://openrouter.ai/qwen/qwen3-max/benchmarks"
+date = "2026-05-30"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 20.5
+metric = "success rate"
+source = "https://openrouter.ai/qwen/qwen3-max/benchmarks"
+date = "2026-05-30"

models/alibaba/qwen3-next-80b-a3b-instruct.toml

@@ -16,3 +16,7 @@ output = 32_768
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct"

models/alibaba/qwen3-next-80b-a3b-thinking.toml

@@ -16,3 +16,7 @@ output = 32_768
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Thinking"

models/alibaba/qwen3.5-122b-a10b.toml

@@ -16,3 +16,13 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3.5-122B-A10B"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 72
+metric = "resolved"
+source = "https://huggingface.co/Qwen/Qwen3.5-122B-A10B"

models/alibaba/qwen3.5-27b.toml

@@ -16,3 +16,13 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3.5-27B"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 72.4
+metric = "resolved"
+source = "https://huggingface.co/Qwen/Qwen3.5-27B"

models/alibaba/qwen3.5-35b-a3b.toml

@@ -16,3 +16,7 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3.5-35B-A3B"

models/alibaba/qwen3.5-397b-a17b.toml

@@ -16,3 +16,13 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3.5-397B-A17B"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 76.4
+metric = "resolved"
+source = "https://huggingface.co/Qwen/Qwen3.5-397B-A17B"

models/alibaba/qwen3.6-27b.toml

@@ -16,3 +16,13 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3.6-27B"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 77.2
+metric = "resolved"
+source = "https://huggingface.co/Qwen/Qwen3.6-27B"

models/alibaba/qwen3.6-35b-a3b.toml

@@ -16,3 +16,13 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3.6-35B-A3B"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 73.4
+metric = "resolved"
+source = "https://huggingface.co/Qwen/Qwen3.6-35B-A3B"

models/alibaba/qwen3.7-plus.toml

@@ -10,9 +10,9 @@ knowledge = "2025-04"
 open_weights = false
 
 [limit]
-context = 131_072
-output = 16_384
+context = 1_000_000
+output = 64_000
 
 [modalities]
-input = ["text"]
+input = ["text", "image"]
 output = ["text"]

models/anthropic/claude-3-5-haiku-20241022.toml

@@ -16,3 +16,10 @@ output = 8_192
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 28.0
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2024-12-21"

models/anthropic/claude-3-5-sonnet-20241022.toml

@@ -16,3 +16,10 @@ output = 8_192
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 51.6
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-01-17"

models/anthropic/claude-3-7-sonnet-20250219.toml

@@ -16,3 +16,10 @@ output = 64_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 64.9
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-02-24"

models/anthropic/claude-fable-5.toml

@@ -0,0 +1,18 @@
+name = "Claude Fable 5"
+family = "claude-fable"
+release_date = "2026-06-09"
+last_updated = "2026-06-09"
+attachment = true
+reasoning = true
+temperature = false
+tool_call = true
+open_weights = false
+knowledge = "2026-01-31"
+
+[limit]
+context = 1_000_000
+output = 128_000
+
+[modalities]
+input = ["text", "image", "pdf"]
+output = ["text"]

models/anthropic/claude-haiku-4-5.toml

@@ -16,3 +16,10 @@ output = 64_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 39.45
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/anthropic/claude-opus-4-0.toml

@@ -16,3 +16,10 @@ output = 32_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 72.0
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-05-25"

models/anthropic/claude-opus-4-20250514.toml

@@ -16,3 +16,10 @@ output = 32_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 72.0
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-05-25"

models/anthropic/claude-opus-4-5-20251101.toml

@@ -16,3 +16,10 @@ output = 64_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 45.89
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/anthropic/claude-opus-4-6.toml

@@ -16,3 +16,79 @@ output = 128_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 51.9
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 33.3
+metric = "score"
+harness = "Claude Code"
+source = "https://labs.scale.com/leaderboard/sweatlas-qna"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 30
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-qna"
+
+[[benchmarks]]
+name = "SWE-Atlas Refactoring"
+score = 35.58
+metric = "score"
+harness = "Claude Code"
+source = "https://labs.scale.com/leaderboard/sweatlas-refactoring"
+
+[[benchmarks]]
+name = "SWE-Atlas Test Writing"
+score = 36.67
+metric = "score"
+harness = "Claude Code"
+source = "https://labs.scale.com/leaderboard/sweatlas-tw"
+
+[[benchmarks]]
+name = "SWE-Atlas Test Writing"
+score = 36.08
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-tw"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Agent Index"
+score = 51.3
+metric = "average pass@1"
+harness = "Claude Code"
+variant = "medium"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 71.9
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 11.8
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+dataset = "hard-aa"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 70.2
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+version = "2.1"
+source = "https://artificialanalysis.ai/agents/coding-agents"

models/anthropic/claude-opus-4-7.toml

@@ -16,3 +16,128 @@ output = 128_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 64.3
+metric = "resolve rate"
+source = "https://www.anthropic.com/news/claude-opus-4-8"
+date = "2026-05-28"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 66.1
+metric = "success rate"
+harness = "Terminus-2"
+version = "2.1"
+source = "https://www.anthropic.com/news/claude-opus-4-8"
+date = "2026-05-28"
+
+[[benchmarks]]
+name = "SWE-Atlas Refactoring"
+score = 48.57
+metric = "score"
+harness = "Claude Code"
+source = "https://labs.scale.com/leaderboard/sweatlas-refactoring"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Agent Index"
+score = 66.6
+metric = "average pass@1"
+harness = "Claude Code"
+variant = "max"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 81
+metric = "pass@1"
+harness = "Claude Code"
+variant = "max"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 44.9
+metric = "pass@1"
+harness = "Claude Code"
+variant = "max"
+dataset = "hard-aa"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 73.8
+metric = "pass@1"
+harness = "Claude Code"
+variant = "max"
+version = "2.1"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Agent Index"
+score = 61.2
+metric = "average pass@1"
+harness = "Cursor CLI"
+variant = "medium"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 78.4
+metric = "pass@1"
+harness = "Cursor CLI"
+variant = "medium"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 34.4
+metric = "pass@1"
+harness = "Cursor CLI"
+variant = "medium"
+dataset = "hard-aa"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 70.6
+metric = "pass@1"
+harness = "Cursor CLI"
+variant = "medium"
+version = "2.1"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Agent Index"
+score = 59.9
+metric = "average pass@1"
+harness = "Claude Code"
+variant = "medium"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 71.7
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 36.4
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+dataset = "hard-aa"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 71.4
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+version = "2.1"
+source = "https://artificialanalysis.ai/agents/coding-agents"

models/anthropic/claude-opus-4-8.toml

@@ -15,3 +15,19 @@ output = 128_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 69.2
+metric = "resolve rate"
+source = "https://www.anthropic.com/news/claude-opus-4-8"
+date = "2026-05-28"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 74.6
+metric = "success rate"
+harness = "Terminus-2"
+version = "2.1"
+source = "https://www.anthropic.com/news/claude-opus-4-8"
+date = "2026-05-28"

models/anthropic/claude-sonnet-4-0.toml

@@ -16,3 +16,17 @@ output = 64_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 61.3
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-05-24"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 42.7
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/anthropic/claude-sonnet-4-20250514.toml

@@ -16,3 +16,10 @@ output = 64_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 61.3
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-05-24"

models/anthropic/claude-sonnet-4-5.toml

@@ -16,3 +16,10 @@ output = 64_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 43.6
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/anthropic/claude-sonnet-4-6.toml

@@ -16,3 +16,58 @@ output = 64_000
 [modalities]
 input = ["text", "image", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 31.2
+metric = "score"
+harness = "Claude Code"
+source = "https://labs.scale.com/leaderboard/sweatlas-qna"
+
+[[benchmarks]]
+name = "SWE-Atlas Refactoring"
+score = 32.21
+metric = "score"
+harness = "Claude Code"
+source = "https://labs.scale.com/leaderboard/sweatlas-refactoring"
+
+[[benchmarks]]
+name = "SWE-Atlas Test Writing"
+score = 31.76
+metric = "score"
+harness = "Claude Code"
+source = "https://labs.scale.com/leaderboard/sweatlas-tw"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Agent Index"
+score = 49.4
+metric = "average pass@1"
+harness = "Claude Code"
+variant = "medium"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 70.3
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 14.9
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+dataset = "hard-aa"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 63.1
+metric = "pass@1"
+harness = "Claude Code"
+variant = "medium"
+version = "2.1"
+source = "https://artificialanalysis.ai/agents/coding-agents"

models/cohere/command-a-03-2025.toml

@@ -16,3 +16,14 @@ output = 8_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/CohereLabs/c4ai-command-a-03-2025"
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 12.0
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-03-14"

models/cohere/command-a-plus-05-2026.toml

@@ -0,0 +1,19 @@
+name = "Command A Plus"
+family = "command-a"
+release_date = "2026-05-20"
+last_updated = "2026-06-09"
+attachment = true
+reasoning = true
+temperature = true
+knowledge = "2025-04-01"
+tool_call = true
+open_weights = true
+structured_output = true
+
+[limit]
+context = 128_000
+output = 64_000
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]

models/cohere/command-r-08-2024.toml

@@ -16,3 +16,7 @@ output = 4_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/CohereLabs/c4ai-command-r-08-2024"

models/cohere/command-r-plus-08-2024.toml

@@ -16,3 +16,7 @@ output = 4_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/CohereLabs/c4ai-command-r-plus-08-2024"

models/cohere/command-r7b-12-2024.toml

@@ -16,3 +16,7 @@ output = 4_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/CohereLabs/c4ai-command-r7b-12-2024"

models/cohere/north-mini-code-1-0.toml

@@ -0,0 +1,19 @@
+name = "North Mini Code"
+family = "north"
+release_date = "2026-06-09"
+last_updated = "2026-06-09"
+attachment = false
+reasoning = true
+temperature = true
+structured_output = true
+knowledge = "2025-09-23"
+tool_call = true
+open_weights = true
+
+[limit]
+context = 256_000
+output = 64_000
+
+[modalities]
+input = ["text"]
+output = ["text"]

models/deepseek/deepseek-chat.toml

@@ -16,3 +16,14 @@ output = 384_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-V3.2"
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 70.2
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-10-03"

models/deepseek/deepseek-r1.toml

@@ -7,7 +7,7 @@ reasoning = true
 temperature = true
 tool_call = true
 knowledge = "2024-07"
-open_weights = false
+open_weights = true
 
 [limit]
 context = 128_000
@@ -16,3 +16,35 @@ output = 32_768
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-R1"
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 56.9
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-01-20"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 15.9
+metric = "index"
+source = "https://openrouter.ai/deepseek/deepseek-r1/benchmarks"
+date = "2026-03-11"
+
+[[benchmarks]]
+name = "SciCode"
+score = 35.7
+metric = "percent correct"
+source = "https://openrouter.ai/deepseek/deepseek-r1/benchmarks"
+date = "2026-03-11"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 6.1
+metric = "success rate"
+source = "https://openrouter.ai/deepseek/deepseek-r1/benchmarks"
+date = "2026-03-11"

models/deepseek/deepseek-reasoner.toml

@@ -16,3 +16,14 @@ output = 384_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-V3.2"
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 74.2
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-10-03"

models/deepseek/deepseek-v4-flash.toml

@@ -17,3 +17,13 @@ output = 384_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 79
+metric = "resolved"
+source = "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash"

models/deepseek/deepseek-v4-pro.toml

@@ -17,3 +17,47 @@ output = 384_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 80.6
+metric = "resolved"
+source = "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Agent Index"
+score = 50.1
+metric = "average pass@1"
+harness = "Claude Code"
+variant = "high"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 67.8
+metric = "pass@1"
+harness = "Claude Code"
+variant = "high"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 18
+metric = "pass@1"
+harness = "Claude Code"
+variant = "high"
+dataset = "hard-aa"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 64.7
+metric = "pass@1"
+harness = "Claude Code"
+variant = "high"
+version = "2.1"
+source = "https://artificialanalysis.ai/agents/coding-agents"

models/google/gemini-2.5-flash-lite.toml

@@ -17,3 +17,24 @@ output = 65_536
 [modalities]
 input = ["text", "image", "audio", "video", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 9.5
+metric = "index"
+source = "https://openrouter.ai/google/gemini-2.5-flash-lite/benchmarks"
+date = "2026-03-11"
+
+[[benchmarks]]
+name = "SciCode"
+score = 19.3
+metric = "percent correct"
+source = "https://openrouter.ai/google/gemini-2.5-flash-lite/benchmarks"
+date = "2026-03-11"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 4.5
+metric = "success rate"
+source = "https://openrouter.ai/google/gemini-2.5-flash-lite/benchmarks"
+date = "2026-03-11"

models/google/gemini-2.5-flash-tts.toml

@@ -0,0 +1,18 @@
+name = "Gemini 2.5 Flash TTS"
+family = "gemini-flash"
+release_date = "2025-09-30"
+last_updated = "2025-12-10"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = false
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 32_768
+output = 16_384
+
+[modalities]
+input = ["text"]
+output = ["audio"]

models/google/gemini-2.5-flash.toml

@@ -17,3 +17,31 @@ output = 65_536
 [modalities]
 input = ["text", "image", "audio", "video", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 55.1
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-05-25"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 22.2
+metric = "index"
+source = "https://openrouter.ai/google/gemini-2.5-flash/benchmarks"
+date = "2026-06-02"
+
+[[benchmarks]]
+name = "SciCode"
+score = 39.4
+metric = "percent correct"
+source = "https://openrouter.ai/google/gemini-2.5-flash/benchmarks"
+date = "2026-06-02"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 13.6
+metric = "success rate"
+source = "https://openrouter.ai/google/gemini-2.5-flash/benchmarks"
+date = "2026-06-02"

models/google/gemini-2.5-pro-tts.toml

@@ -0,0 +1,18 @@
+name = "Gemini 2.5 Pro TTS"
+family = "gemini-pro"
+release_date = "2025-09-30"
+last_updated = "2025-12-10"
+attachment = false
+reasoning = false
+temperature = false
+tool_call = false
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 32_768
+output = 16_384
+
+[modalities]
+input = ["text"]
+output = ["audio"]

models/google/gemini-2.5-pro.toml

@@ -17,3 +17,31 @@ output = 65_536
 [modalities]
 input = ["text", "image", "audio", "video", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 83.1
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-06-06"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 32
+metric = "index"
+source = "https://openrouter.ai/google/gemini-2.5-pro/benchmarks"
+date = "2026-06-02"
+
+[[benchmarks]]
+name = "SciCode"
+score = 42.8
+metric = "percent correct"
+source = "https://openrouter.ai/google/gemini-2.5-pro/benchmarks"
+date = "2026-06-02"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 26.5
+metric = "success rate"
+source = "https://openrouter.ai/google/gemini-2.5-pro/benchmarks"
+date = "2026-06-02"

models/google/gemini-3-flash-preview.toml

@@ -17,3 +17,31 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 34.63
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 8.2
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-qna"
+
+[[benchmarks]]
+name = "SWE-Atlas Refactoring"
+score = 10
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-refactoring"
+
+[[benchmarks]]
+name = "SWE-Atlas Test Writing"
+score = 30.3
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-tw"

models/google/gemini-3-pro-image-preview.toml

@@ -0,0 +1,18 @@
+name = "Nano Banana Pro"
+family = "gemini-pro"
+release_date = "2025-11-20"
+last_updated = "2025-11-20"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = false
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 65_536
+output = 32_768
+
+[modalities]
+input = ["text", "image"]
+output = ["text", "image"]

models/google/gemini-3-pro-preview.toml

@@ -17,3 +17,10 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 43.3
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/google/gemini-3.1-pro-preview.toml

@@ -17,3 +17,81 @@ output = 65_536
 [modalities]
 input = ["text", "image", "video", "audio", "pdf"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 54.2
+metric = "resolve rate"
+source = "https://www.anthropic.com/news/claude-opus-4-8"
+date = "2026-05-28"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 70.3
+metric = "success rate"
+harness = "Terminus-2"
+version = "2.1"
+source = "https://www.anthropic.com/news/claude-opus-4-8"
+date = "2026-05-28"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 46.1
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 13.5
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-qna"
+
+[[benchmarks]]
+name = "SWE-Atlas Refactoring"
+score = 33.81
+metric = "score"
+harness = "Gemini CLI"
+source = "https://labs.scale.com/leaderboard/sweatlas-refactoring"
+
+[[benchmarks]]
+name = "SWE-Atlas Test Writing"
+score = 29.84
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-tw"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Agent Index"
+score = 43
+metric = "average pass@1"
+harness = "Gemini CLI"
+variant = "high"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 45.6
+metric = "pass@1"
+harness = "Gemini CLI"
+variant = "high"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 15.1
+metric = "pass@1"
+harness = "Gemini CLI"
+variant = "high"
+dataset = "hard-aa"
+source = "https://artificialanalysis.ai/agents/coding-agents"
+
+[[benchmarks]]
+name = "Terminal-Bench"
+score = 68.3
+metric = "pass@1"
+harness = "Gemini CLI"
+variant = "high"
+version = "2.1"
+source = "https://artificialanalysis.ai/agents/coding-agents"

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

@@ -16,3 +16,7 @@ output = 32_768
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/google/gemma-4-26B-A4B-it"

models/google/gemma-4-31b-it.toml

@@ -16,3 +16,7 @@ output = 32_768
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/google/gemma-4-31B-it"

models/meta/llama-3.3-70b-instruct.toml

@@ -16,3 +16,28 @@ output = 4_096
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 10.7
+metric = "index"
+source = "https://openrouter.ai/meta-llama/llama-3.3-70b-instruct/benchmarks"
+date = "2026-03-11"
+
+[[benchmarks]]
+name = "SciCode"
+score = 26
+metric = "percent correct"
+source = "https://openrouter.ai/meta-llama/llama-3.3-70b-instruct/benchmarks"
+date = "2026-03-11"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 3
+metric = "success rate"
+source = "https://openrouter.ai/meta-llama/llama-3.3-70b-instruct/benchmarks"
+date = "2026-03-11"

models/meta/llama-4-maverick-17b-instruct.toml

@@ -16,3 +16,21 @@ output = 16_384
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/meta-llama/Llama-4-Maverick-17B-128E-Instruct"
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 15.6
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-04-06"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 5.24
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/meta/llama-4-scout-17b-instruct.toml

@@ -16,3 +16,7 @@ output = 16_384
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct"

models/minimax/MiniMax-M2.1.toml

@@ -15,3 +15,20 @@ output = 131_072
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2.1"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 74
+metric = "resolved"
+source = "https://huggingface.co/MiniMaxAI/MiniMax-M2.1"
+
+[[benchmarks]]
+name = "SWE-Bench Pro"
+score = 36.81
+metric = "resolve rate"
+dataset = "public"
+source = "https://labs.scale.com/leaderboard/swe_bench_pro_public"

models/minimax/MiniMax-M2.5-highspeed.toml

@@ -15,3 +15,7 @@ output = 131_072
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2.5"

models/minimax/MiniMax-M2.5.toml

@@ -15,3 +15,34 @@ output = 131_072
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2.5"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 75.8
+metric = "resolved"
+source = "https://www.swebench.com/"
+
+[[benchmarks]]
+name = "SWE-Atlas Codebase QnA"
+score = 10.3
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-qna"
+
+[[benchmarks]]
+name = "SWE-Atlas Refactoring"
+score = 19.52
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-refactoring"
+
+[[benchmarks]]
+name = "SWE-Atlas Test Writing"
+score = 18.6
+metric = "score"
+harness = "Mini-SWE-Agent"
+source = "https://labs.scale.com/leaderboard/sweatlas-tw"

models/minimax/MiniMax-M2.7-highspeed.toml

@@ -15,3 +15,7 @@ output = 131_072
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"

models/minimax/MiniMax-M2.7.toml

@@ -15,3 +15,7 @@ output = 131_072
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"

models/minimax/MiniMax-M2.toml

@@ -15,3 +15,13 @@ output = 128_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 69.4
+metric = "resolved"
+source = "https://huggingface.co/MiniMaxAI/MiniMax-M2"

models/minimax/MiniMax-M3.toml

@@ -6,7 +6,7 @@ attachment = true
 reasoning = true
 temperature = true
 tool_call = true
-open_weights = true
+open_weights = false
 
 [limit]
 context = 512_000

models/mistral/codestral-latest.toml

@@ -16,3 +16,14 @@ output = 4_096
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Codestral-22B-v0.1"
+
+[[benchmarks]]
+name = "Aider Polyglot"
+score = 11.1
+metric = "percent correct"
+source = "https://aider.chat/docs/leaderboards/"
+date = "2025-01-13"

models/mistral/devstral-2512.toml

@@ -16,3 +16,28 @@ output = 262_144
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 23.7
+metric = "index"
+source = "https://openrouter.ai/mistralai/devstral-2512/benchmarks"
+date = "2026-05-31"
+
+[[benchmarks]]
+name = "SciCode"
+score = 33.1
+metric = "percent correct"
+source = "https://openrouter.ai/mistralai/devstral-2512/benchmarks"
+date = "2026-05-31"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 18.9
+metric = "success rate"
+source = "https://openrouter.ai/mistralai/devstral-2512/benchmarks"
+date = "2026-05-31"

models/mistral/devstral-medium-2507.toml

@@ -7,7 +7,7 @@ reasoning = false
 temperature = true
 tool_call = true
 knowledge = "2025-05"
-open_weights = true
+open_weights = false
 
 [limit]
 context = 128_000
@@ -16,3 +16,10 @@ output = 128_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 61.6
+metric = "resolved"
+source = "https://mistral.ai/news/devstral-2507"
+date = "2025-07-10"

models/mistral/devstral-medium-latest.toml

@@ -16,3 +16,7 @@ output = 262_144
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512"

models/mistral/devstral-small-2507.toml

@@ -16,3 +16,14 @@ output = 128_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Devstral-Small-2507"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 53.6
+metric = "resolved"
+source = "https://mistral.ai/news/devstral-2507"
+date = "2025-07-10"

models/mistral/magistral-medium-latest.toml

@@ -7,7 +7,7 @@ reasoning = true
 temperature = true
 tool_call = true
 knowledge = "2025-06"
-open_weights = true
+open_weights = false
 
 [limit]
 context = 128_000

models/mistral/mistral-large-2411.toml

@@ -16,3 +16,28 @@ output = 16_384
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Large-Instruct-2411"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 13.8
+metric = "index"
+source = "https://openrouter.ai/mistralai/mistral-large-2407/benchmarks"
+date = "2026-03-11"
+
+[[benchmarks]]
+name = "SciCode"
+score = 29.2
+metric = "percent correct"
+source = "https://openrouter.ai/mistralai/mistral-large-2407/benchmarks"
+date = "2026-03-11"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 6.1
+metric = "success rate"
+source = "https://openrouter.ai/mistralai/mistral-large-2407/benchmarks"
+date = "2026-03-11"

models/mistral/mistral-large-2512.toml

@@ -16,3 +16,28 @@ output = 262_144
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 22.7
+metric = "index"
+source = "https://openrouter.ai/mistralai/mistral-large-2512/benchmarks"
+date = "2026-06-04"
+
+[[benchmarks]]
+name = "SciCode"
+score = 36.2
+metric = "percent correct"
+source = "https://openrouter.ai/mistralai/mistral-large-2512/benchmarks"
+date = "2026-06-04"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 15.9
+metric = "success rate"
+source = "https://openrouter.ai/mistralai/mistral-large-2512/benchmarks"
+date = "2026-06-04"

models/mistral/mistral-large-latest.toml

@@ -16,3 +16,7 @@ output = 262_144
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512"

models/mistral/mistral-medium-2505.toml

@@ -16,3 +16,24 @@ output = 131_072
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 13.6
+metric = "index"
+source = "https://openrouter.ai/mistralai/mistral-medium-3/benchmarks"
+date = "2026-05-30"
+
+[[benchmarks]]
+name = "SciCode"
+score = 33.1
+metric = "percent correct"
+source = "https://openrouter.ai/mistralai/mistral-medium-3/benchmarks"
+date = "2026-05-30"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 3.8
+metric = "success rate"
+source = "https://openrouter.ai/mistralai/mistral-medium-3/benchmarks"
+date = "2026-05-30"

models/mistral/mistral-medium-2604.toml

@@ -16,3 +16,13 @@ output = 262_144
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 77.6
+metric = "resolved"
+source = "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B"

models/mistral/mistral-medium-latest.toml

@@ -1,13 +1,15 @@
+# Mistral's live API currently maps this alias to Mistral Medium 3.1,
+# even though Mistral Medium 3.5 is newer and exposed as mistral-medium-2604.
 name = "Mistral Medium (latest)"
 family = "mistral-medium"
-release_date = "2026-04-29"
-last_updated = "2026-04-29"
+release_date = "2025-08-12"
+last_updated = "2025-08-12"
 attachment = true
-reasoning = true
+reasoning = false
 temperature = true
+knowledge = "2025-05"
 tool_call = true
-structured_output = true
-open_weights = true
+open_weights = false
 
 [limit]
 context = 262_144
@@ -16,3 +18,13 @@ output = 262_144
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B"
+
+[[benchmarks]]
+name = "SWE-Bench Verified"
+score = 77.6
+metric = "resolved"
+source = "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B"

models/mistral/mistral-nemo.toml

@@ -16,3 +16,7 @@ output = 128_000
 [modalities]
 input = ["text"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Nemo-Instruct-2407"

models/mistral/mistral-small-2506.toml

@@ -16,3 +16,7 @@ output = 16_384
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506"

models/mistral/mistral-small-2603.toml

@@ -16,3 +16,28 @@ output = 256_000
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603"
+
+[[benchmarks]]
+name = "Artificial Analysis Coding Index"
+score = 24.3
+metric = "index"
+source = "https://openrouter.ai/mistralai/mistral-small-2603/benchmarks"
+date = "2026-06-01"
+
+[[benchmarks]]
+name = "SciCode"
+score = 38
+metric = "percent correct"
+source = "https://openrouter.ai/mistralai/mistral-small-2603/benchmarks"
+date = "2026-06-01"
+
+[[benchmarks]]
+name = "Terminal-Bench Hard"
+score = 17.4
+metric = "success rate"
+source = "https://openrouter.ai/mistralai/mistral-small-2603/benchmarks"
+date = "2026-06-01"

models/mistral/mistral-small-latest.toml

@@ -16,3 +16,7 @@ output = 256_000
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603"

models/mistral/pixtral-12b.toml

@@ -16,3 +16,7 @@ output = 128_000
 [modalities]
 input = ["text", "image"]
 output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Pixtral-12B-2409"