Normalize base-model inheritance

feat(models): add coding benchmarks and weights (#2000 )
* feat(models): add coding benchmarks and weights * feat(models): add more coding benchmarks * feat(models): add agent benchmark scores * feat(models): normalize benchmark metadata
2026-06-04 12:52:03 -05:00 · 2026-06-04 11:09:53 -05:00 · 2026-06-04 11:05:45 -05:00 · 2026-06-04 11:03:53 -05:00 · 2026-06-04 11:31:06 -04:00 · 2026-06-04 00:38:32 -05:00
1679 changed files with 20468 additions and 8199 deletions
@@ -10,6 +10,7 @@ concurrency: ${{ github.workflow }}-${{ github.ref }}

 jobs:
  deploy:
+    if: github.repository == 'anomalyco/models.dev'
    runs-on: ubuntu-latest
    steps:
      - name: Checkout code
@@ -35,3 +36,4 @@ jobs:
      - run: bun sst deploy --stage=dev
        env:
          CLOUDFLARE_API_TOKEN: ${{ secrets.CLOUDFLARE_API_TOKEN }}
+          CLOUDFLARE_DEFAULT_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_DEFAULT_ACCOUNT_ID }}
@@ -2,7 +2,7 @@ name: Sync Model Catalogs

 on:
  schedule:
-    - cron: "17 8 * * *"
+    - cron: "17 * * * *"
  workflow_dispatch:

 permissions:
@@ -13,20 +13,38 @@ permissions:
 concurrency: ${{ github.workflow }}-${{ github.ref }}

 jobs:
+  providers:
+    if: github.repository == 'anomalyco/models.dev'
+    runs-on: ubuntu-latest
+    outputs:
+      matrix: ${{ steps.providers.outputs.matrix }}
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5
+        with:
+          ref: dev
+
+      - name: Setup Bun
+        uses: oven-sh/setup-bun@f4d14e03ff726c06358e5557344e1da148b56cf7
+        with:
+          bun-version: latest
+
+      - name: Install dependencies
+        run: bun install
+
+      - name: List sync providers
+        id: providers
+        run: |
+          matrix="$(bun models:sync --list-providers)"
+          echo "matrix=$matrix" >> "$GITHUB_OUTPUT"
+
  sync:
+    needs: providers
+    if: github.repository == 'anomalyco/models.dev'
    runs-on: ubuntu-latest
    strategy:
      fail-fast: false
-      matrix:
-        include:
-          - group: aggregators
-            title: "chore(sync): update aggregator model catalogs"
-            branch: automation/sync-models-aggregators
-            labels: automation,model-sync,sync-group:aggregators,provider:openrouter
-          - group: cloudflare
-            title: "chore(sync): update cloudflare model catalogs"
-            branch: automation/sync-models-cloudflare
-            labels: automation,model-sync,sync-group:cloudflare,provider:cloudflare-workers-ai
+      matrix: ${{ fromJSON(needs.providers.outputs.matrix) }}

    steps:
      - name: Checkout code
@@ -43,9 +61,13 @@ jobs:
        run: bun install

      - name: Sync model catalogs
-        run: bun models:sync ${{ matrix.group }}
+        run: bun models:sync ${{ matrix.provider }}
        env:
          OPENROUTER_API_KEY: ${{ secrets.OPENROUTER_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 }}
+          XAI_API_KEY: ${{ secrets.XAI_API_KEY }}
          CLOUDFLARE_WORKERS_AI_SYNC_ACCOUNT_ID: ${{ secrets.CLOUDFLARE_WORKERS_AI_SYNC_ACCOUNT_ID }}
          CLOUDFLARE_WORKERS_AI_SYNC_API_TOKEN: ${{ secrets.CLOUDFLARE_WORKERS_AI_SYNC_API_TOKEN }}

@@ -55,9 +77,9 @@ jobs:
      - name: Create pull request
        env:
          GH_TOKEN: ${{ github.token }}
-          BRANCH: ${{ matrix.branch }}
-          LABELS: ${{ matrix.labels }}
-          TITLE: ${{ matrix.title }}
+          BRANCH: automation/sync-models-${{ matrix.provider }}
+          LABELS: automation,model-sync,provider:${{ matrix.provider }}
+          TITLE: "chore(sync): update ${{ matrix.name }} model catalog"
        run: |
          if [ -z "$(git status --porcelain -- providers)" ]; then
            echo "No model catalog changes found."
@@ -66,6 +88,7 @@ jobs:

          git config user.name "github-actions[bot]"
          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 "$BRANCH"
          git add providers
          git commit -m "$TITLE"
@@ -0,0 +1,380 @@
+{
+  "name": ".opencode",
+  "lockfileVersion": 3,
+  "requires": true,
+  "packages": {
+    "": {
+      "dependencies": {
+        "@opencode-ai/plugin": "1.15.13"
+      }
+    },
+    "node_modules/@msgpackr-extract/msgpackr-extract-darwin-arm64": {
+      "version": "3.0.4",
+      "resolved": "https://registry.npmjs.org/@msgpackr-extract/msgpackr-extract-darwin-arm64/-/msgpackr-extract-darwin-arm64-3.0.4.tgz",
+      "integrity": "sha512-LCkGo6JDfaBhgST7UpPWgNgLINpcpabaHfyz5OBx75nUYxBsaEPxjnyNjWpeb/xBup/682QnBfRBy2/LvPutZQ==",
+      "cpu": [
+        "arm64"
+      ],
+      "license": "MIT",
+      "optional": true,
+      "os": [
+        "darwin"
+      ]
+    },
+    "node_modules/@msgpackr-extract/msgpackr-extract-darwin-x64": {
+      "version": "3.0.4",
+      "resolved": "https://registry.npmjs.org/@msgpackr-extract/msgpackr-extract-darwin-x64/-/msgpackr-extract-darwin-x64-3.0.4.tgz",
+      "integrity": "sha512-zExlW9zUJKZH/tOtVMttwjKa4Xm/3KcNjnE3dPN92uCktwavMxpgCA3MoJK/DOnTWsQgo224OaST27/mPNAf+w==",
+      "cpu": [
+        "x64"
+      ],
+      "license": "MIT",
+      "optional": true,
+      "os": [
+        "darwin"
+      ]
+    },
+    "node_modules/@msgpackr-extract/msgpackr-extract-linux-arm": {
+      "version": "3.0.4",
+      "resolved": "https://registry.npmjs.org/@msgpackr-extract/msgpackr-extract-linux-arm/-/msgpackr-extract-linux-arm-3.0.4.tgz",
+      "integrity": "sha512-Tg3yX65f5GbtXLkrYEHE5oibZG9epyYWas7FogTTEJeDEF9JlXJzKgXaNhT3UXlTOeA+AfZpYZYZ0uPj7Cfquw==",
+      "cpu": [
+        "arm"
+      ],
+      "license": "MIT",
+      "optional": true,
+      "os": [
+        "linux"
+      ]
+    },
+    "node_modules/@msgpackr-extract/msgpackr-extract-linux-arm64": {
+      "version": "3.0.4",
+      "resolved": "https://registry.npmjs.org/@msgpackr-extract/msgpackr-extract-linux-arm64/-/msgpackr-extract-linux-arm64-3.0.4.tgz",
+      "integrity": "sha512-dgX0P/9wGPJeHFBG+ZmhgE6bmtMt7NP5CRBGyyktpopdk/mW4POnrpQsSLtKI1dwpc+pPLuXHDh6vvskyQE/sw==",
+      "cpu": [
+        "arm64"
+      ],
+      "license": "MIT",
+      "optional": true,
+      "os": [
+        "linux"
+      ]
+    },
+    "node_modules/@msgpackr-extract/msgpackr-extract-linux-x64": {
+      "version": "3.0.4",
+      "resolved": "https://registry.npmjs.org/@msgpackr-extract/msgpackr-extract-linux-x64/-/msgpackr-extract-linux-x64-3.0.4.tgz",
+      "integrity": "sha512-8TNXMEjJc3QEy7R/x1INhgiU+XakDAFUzBhaz7+Rbrs8NH5UQeHQxxmzsSBJGyV6I1jW79undiQm8tOI+D+8FQ==",
+      "cpu": [
+        "x64"
+      ],
+      "license": "MIT",
+      "optional": true,
+      "os": [
+        "linux"
+      ]
+    },
+    "node_modules/@msgpackr-extract/msgpackr-extract-win32-x64": {
+      "version": "3.0.4",
+      "resolved": "https://registry.npmjs.org/@msgpackr-extract/msgpackr-extract-win32-x64/-/msgpackr-extract-win32-x64-3.0.4.tgz",
+      "integrity": "sha512-CmCXPQrkbwExx3j946/PtHWHbYJiCRBRDl4BlkRQcJB/YOwQxJRTpoo7aTsortjgoJ1x7opzTSxn7C+ASSLVjQ==",
+      "cpu": [
+        "x64"
+      ],
+      "license": "MIT",
+      "optional": true,
+      "os": [
+        "win32"
+      ]
+    },
+    "node_modules/@opencode-ai/plugin": {
+      "version": "1.15.13",
+      "resolved": "https://registry.npmjs.org/@opencode-ai/plugin/-/plugin-1.15.13.tgz",
+      "integrity": "sha512-NFwZGhmxIPijtfz9swPJXDmhOpq4UWP8WjEE7GEMr7FwtJrK/hv6v36nFimed5+OKk+pQCrTJn/vhRW7Io72IA==",
+      "license": "MIT",
+      "dependencies": {
+        "@opencode-ai/sdk": "1.15.13",
+        "effect": "4.0.0-beta.66",
+        "zod": "4.1.8"
+      },
+      "peerDependencies": {
+        "@opentui/core": ">=0.2.16",
+        "@opentui/keymap": ">=0.2.16",
+        "@opentui/solid": ">=0.2.16"
+      },
+      "peerDependenciesMeta": {
+        "@opentui/core": {
+          "optional": true
+        },
+        "@opentui/keymap": {
+          "optional": true
+        },
+        "@opentui/solid": {
+          "optional": true
+        }
+      }
+    },
+    "node_modules/@opencode-ai/sdk": {
+      "version": "1.15.13",
+      "resolved": "https://registry.npmjs.org/@opencode-ai/sdk/-/sdk-1.15.13.tgz",
+      "integrity": "sha512-4TwojIoQ8EG6/mVBuUVYZXiFcwNmiiytEnjnvyuvSJjGwFIlw2YIBFxtSVC3FbwwbwHT63teh1RHiQUUC4U5xw==",
+      "license": "MIT",
+      "dependencies": {
+        "cross-spawn": "7.0.6"
+      }
+    },
+    "node_modules/@standard-schema/spec": {
+      "version": "1.1.0",
+      "resolved": "https://registry.npmjs.org/@standard-schema/spec/-/spec-1.1.0.tgz",
+      "integrity": "sha512-l2aFy5jALhniG5HgqrD6jXLi/rUWrKvqN/qJx6yoJsgKhblVd+iqqU4RCXavm/jPityDo5TCvKMnpjKnOriy0w==",
+      "license": "MIT"
+    },
+    "node_modules/cross-spawn": {
+      "version": "7.0.6",
+      "resolved": "https://registry.npmjs.org/cross-spawn/-/cross-spawn-7.0.6.tgz",
+      "integrity": "sha512-uV2QOWP2nWzsy2aMp8aRibhi9dlzF5Hgh5SHaB9OiTGEyDTiJJyx0uy51QXdyWbtAHNua4XJzUKca3OzKUd3vA==",
+      "license": "MIT",
+      "dependencies": {
+        "path-key": "^3.1.0",
+        "shebang-command": "^2.0.0",
+        "which": "^2.0.1"
+      },
+      "engines": {
+        "node": ">= 8"
+      }
+    },
+    "node_modules/detect-libc": {
+      "version": "2.1.2",
+      "resolved": "https://registry.npmjs.org/detect-libc/-/detect-libc-2.1.2.tgz",
+      "integrity": "sha512-Btj2BOOO83o3WyH59e8MgXsxEQVcarkUOpEYrubB0urwnN10yQ364rsiByU11nZlqWYZm05i/of7io4mzihBtQ==",
+      "license": "Apache-2.0",
+      "optional": true,
+      "engines": {
+        "node": ">=8"
+      }
+    },
+    "node_modules/effect": {
+      "version": "4.0.0-beta.66",
+      "resolved": "https://registry.npmjs.org/effect/-/effect-4.0.0-beta.66.tgz",
+      "integrity": "sha512-4arEr62cziFa8BBVDUwJCJJmaVepXf/kRg7KtC0h8+bufngscrHbwWFhr9c+HonwOF+31U3iD3xUJmw9KzX7Dw==",
+      "license": "MIT",
+      "dependencies": {
+        "@standard-schema/spec": "^1.1.0",
+        "fast-check": "^4.6.0",
+        "find-my-way-ts": "^0.1.6",
+        "ini": "^6.0.0",
+        "kubernetes-types": "^1.30.0",
+        "msgpackr": "^1.11.9",
+        "multipasta": "^0.2.7",
+        "toml": "^4.1.1",
+        "uuid": "^13.0.0",
+        "yaml": "^2.8.3"
+      }
+    },
+    "node_modules/fast-check": {
+      "version": "4.8.0",
+      "resolved": "https://registry.npmjs.org/fast-check/-/fast-check-4.8.0.tgz",
+      "integrity": "sha512-GOJ158CUMnN6cSahsv4+ExARvIDuzzinFjkp0E9WtiBa5zcVeLozVkWaE4IzFcc+Y48Wp1EDlUZsXRyAztQcSg==",
+      "funding": [
+        {
+          "type": "individual",
+          "url": "https://github.com/sponsors/dubzzz"
+        },
+        {
+          "type": "opencollective",
+          "url": "https://opencollective.com/fast-check"
+        }
+      ],
+      "license": "MIT",
+      "dependencies": {
+        "pure-rand": "^8.0.0"
+      },
+      "engines": {
+        "node": ">=12.17.0"
+      }
+    },
+    "node_modules/find-my-way-ts": {
+      "version": "0.1.6",
+      "resolved": "https://registry.npmjs.org/find-my-way-ts/-/find-my-way-ts-0.1.6.tgz",
+      "integrity": "sha512-a85L9ZoXtNAey3Y6Z+eBWW658kO/MwR7zIafkIUPUMf3isZG0NCs2pjW2wtjxAKuJPxMAsHUIP4ZPGv0o5gyTA==",
+      "license": "MIT"
+    },
+    "node_modules/ini": {
+      "version": "6.0.0",
+      "resolved": "https://registry.npmjs.org/ini/-/ini-6.0.0.tgz",
+      "integrity": "sha512-IBTdIkzZNOpqm7q3dRqJvMaldXjDHWkEDfrwGEQTs5eaQMWV+djAhR+wahyNNMAa+qpbDUhBMVt4ZKNwpPm7xQ==",
+      "license": "ISC",
+      "engines": {
+        "node": "^20.17.0 || >=22.9.0"
+      }
+    },
+    "node_modules/isexe": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/isexe/-/isexe-2.0.0.tgz",
+      "integrity": "sha512-RHxMLp9lnKHGHRng9QFhRCMbYAcVpn69smSGcq3f36xjgVVWThj4qqLbTLlq7Ssj8B+fIQ1EuCEGI2lKsyQeIw==",
+      "license": "ISC"
+    },
+    "node_modules/kubernetes-types": {
+      "version": "1.30.0",
+      "resolved": "https://registry.npmjs.org/kubernetes-types/-/kubernetes-types-1.30.0.tgz",
+      "integrity": "sha512-Dew1okvhM/SQcIa2rcgujNndZwU8VnSapDgdxlYoB84ZlpAD43U6KLAFqYo17ykSFGHNPrg0qry0bP+GJd9v7Q==",
+      "license": "Apache-2.0"
+    },
+    "node_modules/msgpackr": {
+      "version": "1.11.12",
+      "resolved": "https://registry.npmjs.org/msgpackr/-/msgpackr-1.11.12.tgz",
+      "integrity": "sha512-RBdJ1Un7yGlXWajrkxcSa93nvQ0w4zBf60c0yYv7YtBelP8H2FA7XsfBbMHtXKXUMUxH7zV3Zuozh+kUQWhHvg==",
+      "license": "MIT",
+      "optionalDependencies": {
+        "msgpackr-extract": "^3.0.2"
+      }
+    },
+    "node_modules/msgpackr-extract": {
+      "version": "3.0.4",
+      "resolved": "https://registry.npmjs.org/msgpackr-extract/-/msgpackr-extract-3.0.4.tgz",
+      "integrity": "sha512-4kmO/MdyUIkLIvTPr8VHLil4AtoKIoniWPIEk5+CDy0xnWC84azhSFmuJ7PxZdsYtiP5kEeQsORAVIeMgxT+Hw==",
+      "hasInstallScript": true,
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "node-gyp-build-optional-packages": "5.2.2"
+      },
+      "bin": {
+        "download-msgpackr-prebuilds": "bin/download-prebuilds.js"
+      },
+      "optionalDependencies": {
+        "@msgpackr-extract/msgpackr-extract-darwin-arm64": "3.0.4",
+        "@msgpackr-extract/msgpackr-extract-darwin-x64": "3.0.4",
+        "@msgpackr-extract/msgpackr-extract-linux-arm": "3.0.4",
+        "@msgpackr-extract/msgpackr-extract-linux-arm64": "3.0.4",
+        "@msgpackr-extract/msgpackr-extract-linux-x64": "3.0.4",
+        "@msgpackr-extract/msgpackr-extract-win32-x64": "3.0.4"
+      }
+    },
+    "node_modules/multipasta": {
+      "version": "0.2.7",
+      "resolved": "https://registry.npmjs.org/multipasta/-/multipasta-0.2.7.tgz",
+      "integrity": "sha512-KPA58d68KgGil15oDqXjkUBEBYc00XvbPj5/X+dyzeo/lWm9Nc25pQRlf1D+gv4OpK7NM0J1odrbu9JNNGvynA==",
+      "license": "MIT"
+    },
+    "node_modules/node-gyp-build-optional-packages": {
+      "version": "5.2.2",
+      "resolved": "https://registry.npmjs.org/node-gyp-build-optional-packages/-/node-gyp-build-optional-packages-5.2.2.tgz",
+      "integrity": "sha512-s+w+rBWnpTMwSFbaE0UXsRlg7hU4FjekKU4eyAih5T8nJuNZT1nNsskXpxmeqSK9UzkBl6UgRlnKc8hz8IEqOw==",
+      "license": "MIT",
+      "optional": true,
+      "dependencies": {
+        "detect-libc": "^2.0.1"
+      },
+      "bin": {
+        "node-gyp-build-optional-packages": "bin.js",
+        "node-gyp-build-optional-packages-optional": "optional.js",
+        "node-gyp-build-optional-packages-test": "build-test.js"
+      }
+    },
+    "node_modules/path-key": {
+      "version": "3.1.1",
+      "resolved": "https://registry.npmjs.org/path-key/-/path-key-3.1.1.tgz",
+      "integrity": "sha512-ojmeN0qd+y0jszEtoY48r0Peq5dwMEkIlCOu6Q5f41lfkswXuKtYrhgoTpLnyIcHm24Uhqx+5Tqm2InSwLhE6Q==",
+      "license": "MIT",
+      "engines": {
+        "node": ">=8"
+      }
+    },
+    "node_modules/pure-rand": {
+      "version": "8.4.0",
+      "resolved": "https://registry.npmjs.org/pure-rand/-/pure-rand-8.4.0.tgz",
+      "integrity": "sha512-IoM8YF/jY0hiugFo/wOWqfmarlE6J0wc6fDK1PhftMk7MGhVZl88sZimmqBBFomLOCSmcCCpsfj7wXASCpvK9A==",
+      "funding": [
+        {
+          "type": "individual",
+          "url": "https://github.com/sponsors/dubzzz"
+        },
+        {
+          "type": "opencollective",
+          "url": "https://opencollective.com/fast-check"
+        }
+      ],
+      "license": "MIT"
+    },
+    "node_modules/shebang-command": {
+      "version": "2.0.0",
+      "resolved": "https://registry.npmjs.org/shebang-command/-/shebang-command-2.0.0.tgz",
+      "integrity": "sha512-kHxr2zZpYtdmrN1qDjrrX/Z1rR1kG8Dx+gkpK1G4eXmvXswmcE1hTWBWYUzlraYw1/yZp6YuDY77YtvbN0dmDA==",
+      "license": "MIT",
+      "dependencies": {
+        "shebang-regex": "^3.0.0"
+      },
+      "engines": {
+        "node": ">=8"
+      }
+    },
+    "node_modules/shebang-regex": {
+      "version": "3.0.0",
+      "resolved": "https://registry.npmjs.org/shebang-regex/-/shebang-regex-3.0.0.tgz",
+      "integrity": "sha512-7++dFhtcx3353uBaq8DDR4NuxBetBzC7ZQOhmTQInHEd6bSrXdiEyzCvG07Z44UYdLShWUyXt5M/yhz8ekcb1A==",
+      "license": "MIT",
+      "engines": {
+        "node": ">=8"
+      }
+    },
+    "node_modules/toml": {
+      "version": "4.1.1",
+      "resolved": "https://registry.npmjs.org/toml/-/toml-4.1.1.tgz",
+      "integrity": "sha512-EBJnVBr3dTXdA89WVFoAIPUqkBjxPMwRqsfuo1r240tKFHXv3zgca4+NJib/h6TyvGF7vOawz0jGuryJCdNHrw==",
+      "license": "MIT",
+      "engines": {
+        "node": ">=20"
+      }
+    },
+    "node_modules/uuid": {
+      "version": "13.0.2",
+      "resolved": "https://registry.npmjs.org/uuid/-/uuid-13.0.2.tgz",
+      "integrity": "sha512-vzi9uRZ926x4XV73S/4qQaTwPXM2JBj6/6lI/byHH1jOpCzb0zDbfytgA9LcN/hzb2l7WQSQnxITOVx5un/wGw==",
+      "funding": [
+        "https://github.com/sponsors/broofa",
+        "https://github.com/sponsors/ctavan"
+      ],
+      "license": "MIT",
+      "bin": {
+        "uuid": "dist-node/bin/uuid"
+      }
+    },
+    "node_modules/which": {
+      "version": "2.0.2",
+      "resolved": "https://registry.npmjs.org/which/-/which-2.0.2.tgz",
+      "integrity": "sha512-BLI3Tl1TW3Pvl70l3yq3Y64i+awpwXqsGBYWkkqMtnbXgrMD+yj7rhW0kuEDxzJaYXGjEW5ogapKNMEKNMjibA==",
+      "license": "ISC",
+      "dependencies": {
+        "isexe": "^2.0.0"
+      },
+      "bin": {
+        "node-which": "bin/node-which"
+      },
+      "engines": {
+        "node": ">= 8"
+      }
+    },
+    "node_modules/yaml": {
+      "version": "2.9.0",
+      "resolved": "https://registry.npmjs.org/yaml/-/yaml-2.9.0.tgz",
+      "integrity": "sha512-2AvhNX3mb8zd6Zy7INTtSpl1F15HW6Wnqj0srWlkKLcpYl/gMIMJiyuGq2KeI2YFxUPjdlB+3Lc10seMLtL4cA==",
+      "license": "ISC",
+      "bin": {
+        "yaml": "bin.mjs"
+      },
+      "engines": {
+        "node": ">= 14.6"
+      },
+      "funding": {
+        "url": "https://github.com/sponsors/eemeli"
+      }
+    },
+    "node_modules/zod": {
+      "version": "4.1.8",
+      "license": "MIT",
+      "funding": {
+        "url": "https://github.com/sponsors/colinhacks"
+      }
+    }
+  }
+}
@@ -31,26 +31,25 @@
 ## Model Configuration

 - Model `id` is **auto-injected** from filename (minus `.toml`) — never put `id` in TOML files
- Models may reuse another model's definition via `extends` (see below); otherwise the full definition must be present in the file
+- Provider models may reuse provider-agnostic facts from `models/` via `base_model`; otherwise the full provider model definition must be present in the file
 - Schema uses `.strict()` — extra fields cause validation errors

-### `[extends]` (inheritance between models)
- Syntax — a table at the top of the TOML:
+### Model metadata and `base_model`
+- Provider-agnostic model facts live under `models/<provider>/<model>.toml`
+- Provider TOMLs can inherit those facts with:
  ```toml
-  [extends]
-  from = "<provider-id>/<model-id>"   # required
-  omit = ["experimental.modes.fast"]  # optional, dot-path strings
+  base_model = "<provider-id>/<model-id>"
+  base_model_omit = ["limit.input"] # optional, dot-path strings
  ```
-  Example: `from = "anthropic/claude-opus-4-6"`
- Resolved at parse time in `generate()`; the final JSON output contains **no** `extends` field — it exists only to cut duplication in the TOMLs
+  Example: `base_model = "anthropic/claude-opus-4-6"`
+- Resolved at parse time in `generate()`; the final provider JSON output contains **no** `base_model` or `base_model_omit` fields
 - Merge semantics:
-  - Plain objects (`[cost]`, `[limit]`, `[modalities]`, `[provider]`, `[experimental]`, …) are **deep-merged**
+  - Plain objects from metadata and provider TOML (`[limit]`, `[modalities]`, …) are **deep-merged**
  - Arrays (e.g. `modalities.input`) and primitives are **replaced** wholesale by the child
-  - Any field the child omits is inherited verbatim from the base
- `omit` runs **after** the merge and deletes each dot-path from the result (used when the child needs to *remove* something the base defines, e.g. a provider-specific experimental mode). Every listed path must exist in the merged model, else an error is thrown. Ancestor tables that become empty as a result are also pruned, so `omit = ["experimental.modes.fast"]` yields no `experimental` key in the final JSON when `fast` was the only mode.
- Chains are allowed (A extends B extends C); cycles throw
- The base model must exist; `[extends.from]` pointing at a missing provider/model is an error
- The `extends` table is stripped before schema validation, so the merged result must still satisfy the strict `Model` schema
+  - Any provider field omitted is inherited verbatim from model metadata
+  - `cost`, `provider`, `experimental`, `reasoning_options`, `interleaved`, and `status` are provider-specific and must be declared in provider TOMLs when needed
+- `base_model_omit` runs **after** the merge and deletes each dot-path from the result. Missing paths are ignored. Ancestor tables that become empty as a result are also pruned.
+- The base model metadata file must exist; `base_model` pointing at a missing `models/` entry is an error

 ### Bedrock Naming Patterns
 - Dated models: `-v1:0` suffix (`anthropic.claude-3-5-sonnet-20241022-v1:0.toml`)
@@ -72,4 +71,4 @@
 | `attachment`, `reasoning`, `tool_call`, `open_weights` | Yes | Boolean capabilities |
 | `cost`, `limit`, `modalities` | Yes | Objects with their own required fields |
 | `family`, `knowledge`, `temperature`, `structured_output` | No | Optional metadata |
-| `status` | No | Use for `"alpha"`, `"beta"`, `"deprecated"` lifecycle |
+| `status` | No | Use for `"alpha"`, `"beta"`, `"deprecated"` lifecycle |
@@ -24,6 +24,18 @@ curl https://models.dev/api.json

 Use the **Model ID** field to do a lookup on any model; it's the identifier used by [AI SDK](https://ai-sdk.dev/).

+Provider-agnostic model metadata is available separately:
+
+```bash
+curl https://models.dev/models.json
+```
+
+Use this for facts about the model itself, independent of where it is served. If you need both provider endpoints and model-only metadata in one response:
+
+```bash
+curl https://models.dev/catalog.json
+```
+
 ### Logos

 Provider logos are available as SVG files:
@@ -40,7 +52,71 @@ The data is stored in the repo as TOML files; organized by provider and model. T

 We need your help keeping the data up to date.

-### Adding a New Model
+### Adding Model Metadata
+
+Model-only facts live in `models/`, using the same path-style IDs as provider models. For example, `models/openai/gpt-5.toml` defines metadata for the underlying GPT-5 model, while `providers/openai/models/gpt-5.toml` defines OpenAI-specific serving details such as pricing.
+
+Use model metadata for provider-agnostic facts:
+
+- `name`, `family`, `release_date`, `last_updated`, `knowledge`
+- `attachment`, `reasoning`, `tool_call`, `structured_output`, `temperature`
+- `[limit]` defaults like context, input, and output token limits
+- `[modalities]` defaults
+- `open_weights`, `license`, `links`, `weights`, and `benchmarks`
+
+Example:
+
+```toml
+name = "GPT-5"
+family = "gpt"
+release_date = "2025-08-07"
+last_updated = "2025-08-07"
+attachment = true
+reasoning = true
+temperature = false
+tool_call = true
+structured_output = true
+open_weights = false
+
+[limit]
+context = 400_000
+input = 272_000
+output = 128_000
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
+
+[[benchmarks]]
+name = "Benchmark Name"
+score = 72.5
+metric = "accuracy"
+source = "https://example.com/results"
+
+[[weights]]
+label = "Model weights"
+url = "https://huggingface.co/example/model"
+format = "safetensors"
+```
+
+Provider TOMLs can inherit these facts with `base_model` and then keep only provider-specific fields or overrides:
+
+```toml
+base_model = "openai/gpt-5"
+
+[cost]
+input = 1.25
+output = 10.00
+cache_read = 0.125
+
+[limit]
+context = 200_000 # optional provider override
+output = 32_000
+```
+
+Provider fields win over model metadata during generation. Use this when the underlying model is the same but a provider serves it with different context limits, modalities, features, or pricing.
+
+### Adding a New Provider Model

 To add a new model, start by checking if the provider already exists in the `providers/` directory. If not, then:

@@ -120,30 +196,31 @@ output = ["text"]           # Supported output modalities
 field = "reasoning_content" # Name of the interleaved field "reasoning_content" or "reasoning_details"
 ```

-#### 3a. Reuse an Existing Model with `extends`
+#### 3a. Reuse Model Metadata with `base_model`

-For wrapper providers that mirror a model from another provider, prefer reusing the canonical model definition instead of duplicating the whole file.
+For wrapper providers that mirror an existing model, prefer referencing the model-only metadata instead of duplicating provider-agnostic fields.

-Use `extends` only for non-first-party wrappers and mirrors. Do not use it inside the actual lab provider directories that act as the canonical source for a model family, for example `providers/anthropic/`, `providers/openai/`, `providers/google/`, `providers/xai/`, `providers/minimax/`, or `providers/moonshot/`.
+Use `base_model` when the provider serves the same underlying model and only provider-specific fields differ.

 ```toml
-[extends]
-from = "anthropic/claude-opus-4-6"
-omit = ["experimental.modes.fast"]
+base_model = "anthropic/claude-opus-4-6"

-[provider]
-npm = "@ai-sdk/anthropic"
+[cost]
+input = 5.00
+output = 25.00
 ```

 Rules:

- `from` must point to another model using `<provider>/<model-id>`.
- `omit` is optional and removes fields after the inherited model and local overrides are merged.
+- `base_model` must point to a TOML file in `models/` using `<provider>/<model-id>`.
 - You can override any top-level model field locally.
 - If you override a nested table like `[cost]`, `[limit]`, or `[modalities]`, include the full values needed for that table.
+- `base_model_omit` is optional and removes inherited model metadata fields after local overrides are merged. Use dot-path strings, for example `base_model_omit = ["limit.input"]`.
 - `id` still comes from the filename; do not add it to the TOML.

-Use `extends` when the wrapper model is materially the same as the source model and only differs by a small set of overrides or omitted fields.
+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

@@ -161,7 +238,7 @@ There's a GitHub Action that will automatically validate your submission against
 - Values are within acceptable ranges
 - TOML syntax is valid

-When converting existing wrapper models to `extends`, compare generated output before and after the change:
+When moving existing provider fields into model metadata, compare generated output before and after the change:

 ```bash
 bun run compare:migrations
@@ -0,0 +1,18 @@
+name = "Qwen Flash"
+family = "qwen"
+release_date = "2025-07-28"
+last_updated = "2025-07-28"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 1_000_000
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,25 @@
+name = "Qwen Max"
+family = "qwen"
+release_date = "2024-04-03"
+last_updated = "2025-01-25"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 32_768
+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"
@@ -0,0 +1,18 @@
+name = "Qwen-Omni Turbo"
+family = "qwen"
+release_date = "2025-01-19"
+last_updated = "2025-03-26"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 32_768
+output = 2_048
+
+[modalities]
+input = ["text", "image", "audio", "video"]
+output = ["text", "audio"]
@@ -0,0 +1,18 @@
+name = "Qwen Plus"
+family = "qwen"
+release_date = "2024-01-25"
+last_updated = "2025-09-11"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 1_000_000
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "Qwen Turbo"
+family = "qwen"
+release_date = "2024-11-01"
+last_updated = "2025-04-28"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 1_000_000
+output = 16_384
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "Qwen-VL Max"
+family = "qwen"
+release_date = "2024-04-08"
+last_updated = "2025-08-13"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 131_072
+output = 8_192
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "Qwen-VL Plus"
+family = "qwen"
+release_date = "2024-01-25"
+last_updated = "2025-08-15"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 131_072
+output = 8_192
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
@@ -0,0 +1,22 @@
+name = "Qwen2.5-VL 72B Instruct"
+family = "qwen"
+release_date = "2024-09"
+last_updated = "2024-09"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = true
+
+[limit]
+context = 131_072
+output = 8_192
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen2.5-VL-72B-Instruct"
@@ -0,0 +1,36 @@
+name = "Qwen3 235B-A22B"
+family = "qwen"
+release_date = "2025-04"
+last_updated = "2025-04"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+
+[limit]
+context = 131_072
+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"
@@ -0,0 +1,29 @@
+name = "Qwen3 32B"
+family = "qwen"
+release_date = "2025-04"
+last_updated = "2025-04"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+
+[limit]
+context = 131_072
+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"
@@ -0,0 +1,43 @@
+name = "Qwen3-Coder 30B-A3B Instruct"
+family = "qwen"
+release_date = "2025-04"
+last_updated = "2025-04"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+
+[limit]
+context = 262_144
+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"
@@ -0,0 +1,29 @@
+name = "Qwen3-Coder 480B-A35B Instruct"
+family = "qwen"
+release_date = "2025-04"
+last_updated = "2025-04"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+
+[limit]
+context = 262_144
+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"
@@ -1,19 +1,16 @@
-name = "GLM 4.5 Air Derestricted Steam"
+name = "Qwen3 Coder Flash"
+family = "qwen"
 release_date = "2025-07-28"
 last_updated = "2025-07-28"
 attachment = false
 reasoning = false
-tool_call = false
-structured_output = false
+temperature = true
+tool_call = true
+knowledge = "2025-04"
 open_weights = false

-[cost]
-input = 0.306
-output = 0.306
-
 [limit]
-context = 220_600
-input = 220_600
+context = 1_000_000
 output = 65_536

 [modalities]
@@ -1,21 +1,17 @@
-name = "Qwen3 Coder 480B TEE"
+name = "Qwen3 Coder Plus"
 family = "qwen"
 release_date = "2025-07-23"
 last_updated = "2025-07-23"
 attachment = false
 reasoning = false
-tool_call = false
-structured_output = false
+temperature = true
+tool_call = true
+knowledge = "2025-04"
 open_weights = false

-[cost]
-input = 1.5
-output = 2.00
-
 [limit]
-context = 128_000
-input = 128_000
-output = 32_768
+context = 1_048_576
+output = 65_536

 [modalities]
 input = ["text"]
@@ -0,0 +1,39 @@
+name = "Qwen3 Max"
+family = "qwen"
+release_date = "2025-09-23"
+last_updated = "2025-09-23"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = false
+
+[limit]
+context = 262_144
+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"
@@ -0,0 +1,22 @@
+name = "Qwen3-Next 80B-A3B Instruct"
+family = "qwen"
+release_date = "2025-09"
+last_updated = "2025-09"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+
+[limit]
+context = 131_072
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Instruct"
@@ -0,0 +1,22 @@
+name = "Qwen3-Next 80B-A3B (Thinking)"
+family = "qwen"
+release_date = "2025-09"
+last_updated = "2025-09"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = true
+
+[limit]
+context = 131_072
+output = 32_768
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3-Next-80B-A3B-Thinking"
@@ -0,0 +1,18 @@
+name = "Qwen3-VL Plus"
+family = "qwen"
+release_date = "2025-09-23"
+last_updated = "2025-09-23"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = false
+
+[limit]
+context = 262_144
+output = 32_768
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
@@ -0,0 +1,28 @@
+name = "Qwen3.5 122B-A10B"
+family = "qwen"
+release_date = "2026-02-23"
+last_updated = "2026-02-23"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 262_144
+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"
@@ -0,0 +1,28 @@
+name = "Qwen3.5 27B"
+family = "qwen"
+release_date = "2026-02-23"
+last_updated = "2026-02-23"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 262_144
+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"
@@ -0,0 +1,22 @@
+name = "Qwen3.5 35B-A3B"
+family = "qwen"
+release_date = "2026-02-23"
+last_updated = "2026-02-23"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 262_144
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video", "audio"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/Qwen/Qwen3.5-35B-A3B"
@@ -0,0 +1,28 @@
+name = "Qwen3.5 397B-A17B"
+family = "qwen"
+release_date = "2026-02-15"
+last_updated = "2026-02-15"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 262_144
+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"
@@ -0,0 +1,18 @@
+name = "Qwen3.5 Plus"
+family = "qwen"
+release_date = "2026-02-16"
+last_updated = "2026-02-16"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = false
+
+[limit]
+context = 1_000_000
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]
@@ -0,0 +1,28 @@
+name = "Qwen3.6 27B"
+family = "qwen"
+release_date = "2026-04-22"
+last_updated = "2026-04-22"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 262_144
+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"
@@ -0,0 +1,28 @@
+name = "Qwen3.6 35B-A3B"
+family = "qwen"
+release_date = "2026-04-17"
+last_updated = "2026-04-17"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 262_144
+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"
@@ -0,0 +1,18 @@
+name = "Qwen3.6 Flash"
+family = "qwen3.6"
+release_date = "2026-04-27"
+last_updated = "2026-04-27"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = false
+
+[limit]
+context = 1_000_000
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "Qwen3.6 Max Preview"
+family = "qwen"
+release_date = "2026-04-20"
+last_updated = "2026-04-20"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = false
+
+[limit]
+context = 262_144
+output = 65_536
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "Qwen3.6 Plus"
+family = "qwen"
+release_date = "2026-04-02"
+last_updated = "2026-04-02"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = false
+
+[limit]
+context = 1_000_000
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]
@@ -0,0 +1,17 @@
+name = "Qwen3.7 Max"
+family = "qwen"
+release_date = "2026-05-21"
+last_updated = "2026-05-21"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = false
+
+[limit]
+context = 1_000_000
+output = 65_536
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "Qwen3.7 Plus"
+family = "qwen"
+release_date = "2026-06-02"
+last_updated = "2026-06-02"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-04"
+open_weights = false
+
+[limit]
+context = 131_072
+output = 16_384
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "QwQ Plus"
+family = "qwen"
+release_date = "2025-03-05"
+last_updated = "2025-03-05"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2024-04"
+open_weights = false
+
+[limit]
+context = 131_072
+output = 8_192
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,25 @@
+name = "Claude Haiku 3.5"
+family = "claude-haiku"
+release_date = "2024-10-22"
+last_updated = "2024-10-22"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-07-31"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -0,0 +1,25 @@
+name = "Claude Sonnet 3.5 v2"
+family = "claude-sonnet"
+release_date = "2024-10-22"
+last_updated = "2024-10-22"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-04-30"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -0,0 +1,25 @@
+name = "Claude Sonnet 3.7"
+family = "claude-sonnet"
+release_date = "2025-02-19"
+last_updated = "2025-02-19"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2024-10-31"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -1,19 +1,16 @@
-name = "Claude 3.7 Sonnet Thinking (8K)"
-release_date = "2025-02-24"
-last_updated = "2025-02-24"
+name = "Claude Haiku 4.5"
+family = "claude-haiku"
+release_date = "2025-10-15"
+last_updated = "2025-10-15"
 attachment = true
 reasoning = true
+temperature = true
 tool_call = true
-structured_output = true
+knowledge = "2025-02-28"
 open_weights = false

-[cost]
-input = 2.992
-output = 14.994
-
 [limit]
 context = 200_000
-input = 200_000
 output = 64_000

 [modalities]
@@ -0,0 +1,25 @@
+name = "Claude Haiku 4.5 (latest)"
+family = "claude-haiku"
+release_date = "2025-10-15"
+last_updated = "2025-10-15"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-02-28"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -0,0 +1,25 @@
+name = "Claude Opus 4 (latest)"
+family = "claude-opus"
+release_date = "2025-05-22"
+last_updated = "2025-05-22"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-03-31"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -1,23 +1,18 @@
 name = "Claude Opus 4.1"
 family = "claude-opus"
-status = "deprecated"
 release_date = "2025-08-05"
 last_updated = "2025-08-05"
 attachment = true
 reasoning = true
 temperature = true
-tool_call = false
+tool_call = true
 knowledge = "2025-03-31"
 open_weights = false

-[cost]
-input = 0
-output = 0
-
 [limit]
-context = 80_000
-output = 16_000
+context = 200_000
+output = 32_000

 [modalities]
-input = ["text", "image"]
+input = ["text", "image", "pdf"]
 output = ["text"]
@@ -0,0 +1,18 @@
+name = "Claude Opus 4.1 (latest)"
+family = "claude-opus"
+release_date = "2025-08-05"
+last_updated = "2025-08-05"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-03-31"
+open_weights = false
+
+[limit]
+context = 200_000
+output = 32_000
+
+[modalities]
+input = ["text", "image", "pdf"]
+output = ["text"]
@@ -0,0 +1,25 @@
+name = "Claude Opus 4"
+family = "claude-opus"
+release_date = "2025-05-22"
+last_updated = "2025-05-22"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-03-31"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -0,0 +1,25 @@
+name = "Claude Opus 4.5"
+family = "claude-opus"
+release_date = "2025-11-01"
+last_updated = "2025-11-01"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-03-31"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -1,19 +1,16 @@
-name = "Claude 3.7 Sonnet Thinking (1K)"
-release_date = "2025-02-24"
-last_updated = "2025-02-24"
+name = "Claude Opus 4.5 (latest)"
+family = "claude-opus"
+release_date = "2025-11-24"
+last_updated = "2025-11-24"
 attachment = true
 reasoning = true
+temperature = true
 tool_call = true
-structured_output = true
+knowledge = "2025-03-31"
 open_weights = false

-[cost]
-input = 2.992
-output = 14.994
-
 [limit]
 context = 200_000
-input = 200_000
 output = 64_000

 [modalities]
@@ -0,0 +1,94 @@
+name = "Claude Opus 4.6"
+family = "claude-opus"
+release_date = "2026-02-05"
+last_updated = "2026-03-13"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-05-31"
+open_weights = false
+
+[limit]
+context = 1_000_000
+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"
@@ -0,0 +1,143 @@
+name = "Claude Opus 4.7"
+family = "claude-opus"
+release_date = "2026-04-16"
+last_updated = "2026-04-16"
+attachment = true
+reasoning = true
+temperature = false
+tool_call = true
+knowledge = "2026-01-31"
+open_weights = false
+
+[limit]
+context = 1_000_000
+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"
@@ -0,0 +1,33 @@
+name = "Claude Opus 4.8"
+family = "claude-opus"
+release_date = "2026-05-28"
+last_updated = "2026-05-28"
+attachment = true
+reasoning = true
+temperature = false
+tool_call = true
+open_weights = false
+
+[limit]
+context = 1_000_000
+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"
@@ -0,0 +1,32 @@
+name = "Claude Sonnet 4 (latest)"
+family = "claude-sonnet"
+release_date = "2025-05-22"
+last_updated = "2025-05-22"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-03-31"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -0,0 +1,25 @@
+name = "Claude Sonnet 4"
+family = "claude-sonnet"
+release_date = "2025-05-22"
+last_updated = "2025-05-22"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-03-31"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -1,19 +1,16 @@
-name = "Claude 3.7 Sonnet Thinking (32K)"
-release_date = "2025-07-15"
-last_updated = "2025-07-15"
+name = "Claude Sonnet 4.5"
+family = "claude-sonnet"
+release_date = "2025-09-29"
+last_updated = "2025-09-29"
 attachment = true
 reasoning = true
+temperature = true
 tool_call = true
-structured_output = true
+knowledge = "2025-07-31"
 open_weights = false

-[cost]
-input = 2.992
-output = 14.994
-
 [limit]
 context = 200_000
-input = 200_000
 output = 64_000

 [modalities]
@@ -0,0 +1,25 @@
+name = "Claude Sonnet 4.5 (latest)"
+family = "claude-sonnet"
+release_date = "2025-09-29"
+last_updated = "2025-09-29"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-07-31"
+open_weights = false
+
+[limit]
+context = 200_000
+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"
@@ -0,0 +1,73 @@
+name = "Claude Sonnet 4.6"
+family = "claude-sonnet"
+release_date = "2026-02-17"
+last_updated = "2026-03-13"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-08-31"
+open_weights = false
+
+[limit]
+context = 1_000_000
+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"
@@ -0,0 +1,29 @@
+name = "Command A"
+family = "command-a"
+release_date = "2025-03-13"
+last_updated = "2025-03-13"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-06-01"
+open_weights = true
+
+[limit]
+context = 256_000
+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"
@@ -0,0 +1,22 @@
+name = "Command R"
+family = "command-r"
+release_date = "2024-08-30"
+last_updated = "2024-08-30"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-06-01"
+open_weights = true
+
+[limit]
+context = 128_000
+output = 4_000
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/CohereLabs/c4ai-command-r-08-2024"
@@ -0,0 +1,22 @@
+name = "Command R+"
+family = "command-r"
+release_date = "2024-08-30"
+last_updated = "2024-08-30"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-06-01"
+open_weights = true
+
+[limit]
+context = 128_000
+output = 4_000
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/CohereLabs/c4ai-command-r-plus-08-2024"
@@ -0,0 +1,22 @@
+name = "Command R7B"
+family = "command-r"
+release_date = "2024-02-27"
+last_updated = "2024-02-27"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-06-01"
+open_weights = true
+
+[limit]
+context = 128_000
+output = 4_000
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/CohereLabs/c4ai-command-r7b-12-2024"
@@ -0,0 +1,29 @@
+name = "DeepSeek Chat"
+family = "deepseek"
+release_date = "2025-12-01"
+last_updated = "2026-02-28"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-09"
+open_weights = true
+
+[limit]
+context = 1_000_000
+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"
@@ -0,0 +1,50 @@
+name = "DeepSeek-R1"
+family = "deepseek-thinking"
+release_date = "2025-01-20"
+last_updated = "2025-05-29"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2024-07"
+open_weights = true
+
+[limit]
+context = 128_000
+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"
@@ -0,0 +1,29 @@
+name = "DeepSeek Reasoner"
+family = "deepseek-thinking"
+release_date = "2025-12-01"
+last_updated = "2026-02-28"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-09"
+open_weights = true
+
+[limit]
+context = 1_000_000
+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"
@@ -0,0 +1,29 @@
+name = "DeepSeek V4 Flash"
+family = "deepseek-flash"
+release_date = "2026-04-24"
+last_updated = "2026-04-24"
+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"]
+
+[[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"
@@ -0,0 +1,63 @@
+name = "DeepSeek V4 Pro"
+family = "deepseek-thinking"
+release_date = "2026-04-24"
+last_updated = "2026-04-24"
+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"]
+
+[[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"
@@ -0,0 +1,19 @@
+name = "Gemini 2.0 Flash-Lite"
+family = "gemini-flash-lite"
+release_date = "2024-12-11"
+last_updated = "2024-12-11"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2024-06"
+open_weights = false
+
+[limit]
+context = 1_048_576
+output = 8_192
+
+[modalities]
+input = ["text", "image", "audio", "video", "pdf"]
+output = ["text"]
@@ -0,0 +1,19 @@
+name = "Gemini 2.0 Flash"
+family = "gemini-flash"
+release_date = "2024-12-11"
+last_updated = "2024-12-11"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2024-06"
+open_weights = false
+
+[limit]
+context = 1_048_576
+output = 8_192
+
+[modalities]
+input = ["text", "image", "audio", "video", "pdf"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "Nano Banana"
+family = "gemini-flash"
+release_date = "2025-08-26"
+last_updated = "2025-08-26"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = false
+knowledge = "2025-06"
+open_weights = false
+
+[limit]
+context = 32_768
+output = 32_768
+
+[modalities]
+input = ["text", "image"]
+output = ["text", "image"]
@@ -0,0 +1,40 @@
+name = "Gemini 2.5 Flash-Lite"
+family = "gemini-flash-lite"
+release_date = "2025-06-17"
+last_updated = "2025-06-17"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+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"
@@ -0,0 +1,47 @@
+name = "Gemini 2.5 Flash"
+family = "gemini-flash"
+release_date = "2025-03-20"
+last_updated = "2025-06-05"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+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"
@@ -0,0 +1,47 @@
+name = "Gemini 2.5 Pro"
+family = "gemini-pro"
+release_date = "2025-03-20"
+last_updated = "2025-06-05"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+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"
@@ -0,0 +1,47 @@
+name = "Gemini 3 Flash Preview"
+family = "gemini-flash"
+release_date = "2025-12-17"
+last_updated = "2025-12-17"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+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"
@@ -1,6 +1,5 @@
 name = "Gemini 3 Pro Preview"
 family = "gemini-pro"
-status = "deprecated"
 release_date = "2025-11-18"
 last_updated = "2025-11-18"
 attachment = true
@@ -11,15 +10,17 @@ structured_output = true
 knowledge = "2025-01"
 open_weights = false

-[cost]
-input = 0
-output = 0
-
 [limit]
-context = 128_000
-output = 64_000
-input = 128_000
+context = 1_048_576
+output = 65_536

 [modalities]
-input = ["text", "image", "audio", "video"]
+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"
@@ -0,0 +1,18 @@
+name = "Nano Banana 2"
+family = "gemini-flash"
+release_date = "2026-02-26"
+last_updated = "2026-02-26"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = false
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 65_536
+output = 65_536
+
+[modalities]
+input = ["text", "image", "pdf"]
+output = ["text", "image"]
@@ -0,0 +1,19 @@
+name = "Gemini 3.1 Flash Lite Preview"
+family = "gemini-flash-lite"
+release_date = "2026-03-03"
+last_updated = "2026-03-03"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video", "audio", "pdf"]
+output = ["text"]
@@ -0,0 +1,19 @@
+name = "Gemini 3.1 Flash Lite"
+family = "gemini-flash-lite"
+release_date = "2026-05-07"
+last_updated = "2026-05-07"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video", "audio", "pdf"]
+output = ["text"]
@@ -0,0 +1,19 @@
+name = "Gemini 3.1 Pro Preview Custom Tools"
+family = "gemini-pro"
+release_date = "2026-02-19"
+last_updated = "2026-02-19"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video", "audio", "pdf"]
+output = ["text"]
@@ -0,0 +1,97 @@
+name = "Gemini 3.1 Pro Preview"
+family = "gemini-pro"
+release_date = "2026-02-19"
+last_updated = "2026-02-19"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+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"
@@ -0,0 +1,19 @@
+name = "Gemini 3.5 Flash"
+family = "gemini-flash"
+release_date = "2026-05-19"
+last_updated = "2026-05-19"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+output = 65_536
+
+[modalities]
+input = ["text", "image", "video", "audio", "pdf"]
+output = ["text"]
@@ -0,0 +1,18 @@
+name = "Gemini Embedding 001"
+family = "gemini"
+release_date = "2025-05-20"
+last_updated = "2025-05-20"
+attachment = false
+reasoning = false
+temperature = false
+tool_call = false
+knowledge = "2025-05"
+open_weights = false
+
+[limit]
+context = 2_048
+output = 1
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,19 @@
+name = "Gemini Flash Latest"
+family = "gemini-flash"
+release_date = "2025-09-25"
+last_updated = "2025-09-25"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+output = 65_536
+
+[modalities]
+input = ["text", "image", "audio", "video", "pdf"]
+output = ["text"]
@@ -0,0 +1,19 @@
+name = "Gemini Flash-Lite Latest"
+family = "gemini-flash-lite"
+release_date = "2025-09-25"
+last_updated = "2025-09-25"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+knowledge = "2025-01"
+open_weights = false
+
+[limit]
+context = 1_048_576
+output = 65_536
+
+[modalities]
+input = ["text", "image", "audio", "video", "pdf"]
+output = ["text"]
@@ -0,0 +1,22 @@
+name = "Gemma 4 26B A4B IT"
+family = "gemma"
+release_date = "2026-04-02"
+last_updated = "2026-04-02"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 262_144
+output = 32_768
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/google/gemma-4-26B-A4B-it"
@@ -0,0 +1,22 @@
+name = "Gemma 4 31B IT"
+family = "gemma"
+release_date = "2026-04-02"
+last_updated = "2026-04-02"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+structured_output = true
+open_weights = true
+
+[limit]
+context = 262_144
+output = 32_768
+
+[modalities]
+input = ["text", "image"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/google/gemma-4-31B-it"
@@ -0,0 +1,43 @@
+name = "Llama-3.3-70B-Instruct"
+family = "llama"
+release_date = "2024-12-06"
+last_updated = "2024-12-06"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2023-12"
+open_weights = true
+
+[limit]
+context = 128_000
+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"
@@ -0,0 +1,36 @@
+name = "Llama 4 Maverick 17B Instruct"
+family = "llama"
+release_date = "2025-04-05"
+last_updated = "2025-04-05"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-08"
+open_weights = true
+
+[limit]
+context = 1_000_000
+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"
@@ -0,0 +1,22 @@
+name = "Llama 4 Scout 17B Instruct"
+family = "llama"
+release_date = "2025-04-05"
+last_updated = "2025-04-05"
+attachment = true
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-08"
+open_weights = true
+
+[limit]
+context = 3_500_000
+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"
@@ -0,0 +1,34 @@
+name = "MiniMax-M2.1"
+family = "minimax"
+release_date = "2025-12-23"
+last_updated = "2025-12-23"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = true
+
+[limit]
+context = 204_800
+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"
@@ -0,0 +1,21 @@
+name = "MiniMax-M2.5-highspeed"
+family = "minimax"
+release_date = "2026-02-13"
+last_updated = "2026-02-13"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = true
+
+[limit]
+context = 204_800
+output = 131_072
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2.5"
@@ -0,0 +1,48 @@
+name = "MiniMax-M2.5"
+family = "minimax"
+release_date = "2026-02-12"
+last_updated = "2026-02-12"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = true
+
+[limit]
+context = 204_800
+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"
@@ -0,0 +1,21 @@
+name = "MiniMax-M2.7-highspeed"
+family = "minimax"
+release_date = "2026-03-18"
+last_updated = "2026-03-18"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = true
+
+[limit]
+context = 204_800
+output = 131_072
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
@@ -0,0 +1,21 @@
+name = "MiniMax-M2.7"
+family = "minimax"
+release_date = "2026-03-18"
+last_updated = "2026-03-18"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = true
+
+[limit]
+context = 204_800
+output = 131_072
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
@@ -0,0 +1,27 @@
+name = "MiniMax-M2"
+family = "minimax"
+release_date = "2025-10-27"
+last_updated = "2025-10-27"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = true
+
+[limit]
+context = 196_608
+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"
@@ -0,0 +1,17 @@
+name = "MiniMax-M3"
+family = "minimax"
+release_date = "2026-06-01"
+last_updated = "2026-06-01"
+attachment = true
+reasoning = true
+temperature = true
+tool_call = true
+open_weights = false
+
+[limit]
+context = 512_000
+output = 128_000
+
+[modalities]
+input = ["text", "image", "video"]
+output = ["text"]
@@ -0,0 +1,29 @@
+name = "Codestral (latest)"
+family = "codestral"
+release_date = "2024-05-29"
+last_updated = "2025-01-04"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-10"
+open_weights = true
+
+[limit]
+context = 256_000
+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"
@@ -0,0 +1,43 @@
+name = "Devstral 2"
+family = "devstral"
+release_date = "2025-12-09"
+last_updated = "2025-12-09"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-12"
+open_weights = true
+
+[limit]
+context = 262_144
+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"
@@ -0,0 +1,25 @@
+name = "Devstral Medium"
+family = "devstral"
+release_date = "2025-07-10"
+last_updated = "2025-07-10"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-05"
+open_weights = false
+
+[limit]
+context = 128_000
+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"
@@ -0,0 +1,22 @@
+name = "Devstral 2 (latest)"
+family = "devstral"
+release_date = "2025-12-02"
+last_updated = "2025-12-02"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-12"
+open_weights = true
+
+[limit]
+context = 262_144
+output = 262_144
+
+[modalities]
+input = ["text"]
+output = ["text"]
+
+[[weights]]
+label = "Hugging Face"
+url = "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512"
@@ -0,0 +1,29 @@
+name = "Devstral Small"
+family = "devstral"
+release_date = "2025-07-10"
+last_updated = "2025-07-10"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2025-05"
+open_weights = true
+
+[limit]
+context = 128_000
+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"
@@ -0,0 +1,18 @@
+name = "Magistral Medium (latest)"
+family = "magistral-medium"
+release_date = "2025-03-17"
+last_updated = "2025-03-20"
+attachment = false
+reasoning = true
+temperature = true
+tool_call = true
+knowledge = "2025-06"
+open_weights = false
+
+[limit]
+context = 128_000
+output = 16_384
+
+[modalities]
+input = ["text"]
+output = ["text"]
@@ -0,0 +1,43 @@
+name = "Mistral Large 2.1"
+family = "mistral-large"
+release_date = "2024-11-01"
+last_updated = "2024-11-04"
+attachment = false
+reasoning = false
+temperature = true
+tool_call = true
+knowledge = "2024-11"
+open_weights = true
+
+[limit]
+context = 131_072
+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"
--- a/Show More
+++ b/Show More