@grida/ai-models 1.20261105.0 → 1.20261105.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/PROVENANCE.md CHANGED
@@ -1282,3 +1282,132 @@ service contracts, not claims of parameter or pricing parity with another
1282
1282
  provider. Reference limits, dimension rules and supported video options are
1283
1283
  defined by the linked operation-specific schemas. No native model release
1284
1284
  dates or other providers' listed prices change.
1285
+
1286
+ ## November 2026 family additions
1287
+
1288
+ ### `openai/gpt-6.1-terra`
1289
+
1290
+ OpenAI's October 26 changelog establishes broad availability of the exact model.
1291
+ The native model page supplies the 1,050,000-token context, 128,000-token output
1292
+ limit, a separate 922,000-token maximum input, supported reasoning efforts and Standard token rates. Thirty-minute cache
1293
+ writes are a separate meter. The request-wide long-context boundary is strictly
1294
+ more than 272,000 input tokens on OpenAI and Vercel; OpenRouter's feed instead
1295
+ starts its override at 272,000. Each binding retains its own boundary.
1296
+
1297
+ - https://developers.openai.com/api/docs/changelog
1298
+ - https://developers.openai.com/api/docs/models/gpt-6.1-terra
1299
+ - https://vercel.com/ai-gateway/models/gpt-6.1-terra
1300
+ - https://ai-gateway.vercel.sh/v1/models
1301
+ - https://openrouter.ai/api/v1/models
1302
+
1303
+ Tool calling on the native API requires Responses. Image input is supported, but
1304
+ the image-tokenization tables do not identify this exact model; no multiplier is
1305
+ borrowed from Sol or a previous Terra model.
1306
+ https://developers.openai.com/api/docs/guides/images-vision
1307
+
1308
+ ### `google/gemini-4-argon`
1309
+
1310
+ The October 13 changelog establishes general availability, following an earlier
1311
+ restricted rollout. Native Standard regular rates are $4 input, $20 output and
1312
+ $0.20 implicit-cache reads per million tokens. The native pricing page explicitly
1313
+ publishes those regular rates. Vercel's model-page backing-model records independently
1314
+ publish `inputListCostTiers`, `outputListCostTiers` and
1315
+ `cachedInputListCostTiers` with those amounts and `priceType: "list"` for its
1316
+ Google and Vertex offerings. The catalogue records these list amounts rather than
1317
+ the introductory rates in current model feeds. No rate is reconstructed from a
1318
+ discount percentage. OpenRouter's verified binding remains without a regular
1319
+ price because its feed does not separately establish one.
1320
+
1321
+ - https://ai.google.dev/gemini-api/docs/changelog
1322
+ - https://ai.google.dev/gemini-api/docs/pricing
1323
+ - https://vercel.com/ai-gateway/models/gemini-4-argon
1324
+ - https://openrouter.ai/api/v1/models
1325
+
1326
+ The native input, thinking and output share a 1,048,576-token window. Default
1327
+ synchronous output is 65,536; larger generations use continuation hops or
1328
+ background execution. The native card records that default output ceiling, while
1329
+ provider discovery limits retain their published values with this qualification.
1330
+ The media-resolution table explicitly covers Gemini 4 and Gemini 3 and marks its
1331
+ image-token counts as approximate. Explicit caching is not supported.
1332
+
1333
+ - https://ai.google.dev/gemini-api/docs/models/gemini-4-argon
1334
+ - https://ai.google.dev/gemini-api/docs/latest-model
1335
+ - https://ai.google.dev/gemini-api/docs/media-resolution
1336
+
1337
+ ### `openai/gpt-image-2.6-flare` and `openai/gpt-image-2.6-sunburst`
1338
+
1339
+ The November 4 API changelog establishes release of both variants. Native token
1340
+ rates remain distinct for text input, image input and image output. fal separately
1341
+ publishes its text-output meter and rounds total cost upward to $0.0001.
1342
+
1343
+ - https://developers.openai.com/api/docs/changelog
1344
+ - https://developers.openai.com/api/docs/models/gpt-image-2.6-flare
1345
+ - https://developers.openai.com/api/docs/models/gpt-image-2.6-sunburst
1346
+ - https://fal.ai/models/openai/gpt-image-2.6/flare/text-to-image
1347
+ - https://fal.ai/models/openai/gpt-image-2.6/sunburst/text-to-image
1348
+
1349
+ The current official output calculator uses the same base grid axes as GPT Image
1350
+ 2.5, then multiplies the **already-ceiled** base token count by 4.2 for `low`, 3.3
1351
+ for `medium`, and 1 for the other qualities. Its final JavaScript `Math.round`
1352
+ step rounds positive half ties upward. This second step is recorded explicitly;
1353
+ copying the 2.5 formula alone would understate low and medium output tokens. Square
1354
+ examples are reproduced from the published calculator, not inferred from rounded
1355
+ image-price examples. The calculator estimates raster output only; it excludes
1356
+ text/image input and partial-image charges.
1357
+
1358
+ - https://developers.openai.com/api/docs/guides/image-generation
1359
+ - https://developers.openai.com/api/reference/resources/images/methods/generate
1360
+ - https://developers.openai.com/_astro/GptImageTokenCalculator.react.SM_UvXcs.js
1361
+
1362
+ Provider format differences are intentional. The native guide and exact API
1363
+ references document GIF on generation and editing, with 2–16 frames. Native GIF
1364
+ output requires access to the selected model's GIF preview; documentation alone
1365
+ does not establish account eligibility. GIF cannot be combined with masks,
1366
+ streaming, partial images or custom output compression. The guide documents
1367
+ 20–2,000 ms frame durations rounded to 10 ms; the API reference separately
1368
+ explains encoded-delay behavior. All frames/outputs share the request pixel
1369
+ budget. fal generation permits GIF and SVG, whereas its edit schema lists GIF
1370
+ but no SVG, also with 2–16 frames. OpenRouter lists PNG/JPEG/WebP only. Each
1371
+ binding preserves its own documented surface without inferring access or parity.
1372
+ SVG includes separately billed authoring-model tokens, not just raster token
1373
+ rates. Transparent backgrounds require PNG or WebP.
1374
+
1375
+ - https://developers.openai.com/api/reference/resources/images/methods/edit
1376
+
1377
+ - https://fal.ai/models/openai/gpt-image-2.6/flare/text-to-image/api
1378
+ - https://fal.ai/models/openai/gpt-image-2.6/flare/edit/api
1379
+ - https://fal.ai/models/openai/gpt-image-2.6/sunburst/text-to-image/api
1380
+ - https://fal.ai/models/openai/gpt-image-2.6/sunburst/edit/api
1381
+ - https://openrouter.ai/api/v1/images/models/openai/gpt-image-2.6-flare/endpoints
1382
+ - https://openrouter.ai/api/v1/images/models/openai/gpt-image-2.6-sunburst/endpoints
1383
+
1384
+ ### `bfl/flux-3-image`
1385
+
1386
+ The earlier FLUX 3 announcement describes image access as forthcoming, not broad
1387
+ availability. A durable native launch date is not established by the current
1388
+ pricing page. The recorded October 5 date therefore uses `provider_endpoint`
1389
+ basis and Vercel's explicitly dated listing; it does not claim the model first
1390
+ became broadly available on that day.
1391
+
1392
+ - https://bfl.ai/pricing
1393
+ - https://bfl.ai/blog/flux-3
1394
+ - https://vercel.com/ai-gateway/models/flux-3-image
1395
+
1396
+ Native, Vercel and OpenRouter publish per-image resolution tiers. The native price
1397
+ includes reference images and prompts at no extra charge. fal's page and billing
1398
+ metadata instead name a $0.048 megapixel unit; that source does not establish
1399
+ quantity rounding or all resolution-tier rates. The edit page explicitly states
1400
+ one flat price per image regardless of reference count, establishing that
1401
+ references add no separate charge. This does not establish identical prices
1402
+ across resolutions, and missing rates are not silently replaced with BFL's flat
1403
+ per-image tiers. fal's exact API schema
1404
+ lists `768sq`, `1k`, `1.5k`, `2k`, `4k`; its narrative table instead lists `512sq`
1405
+ and omits `1.5k`. The exact endpoint schema supplies the fal parameter list.
1406
+
1407
+ - https://bfl.ai/pricing?category=flux+3+image
1408
+ - https://ai-gateway.vercel.sh/v1/models
1409
+ - https://openrouter.ai/api/v1/images/models/black-forest-labs/flux-3-image/endpoints
1410
+ - https://fal.ai/models/blackforestlabs/flux-3/text-to-image
1411
+ - https://fal.ai/models/blackforestlabs/flux-3/text-to-image/api
1412
+ - https://fal.ai/models/blackforestlabs/flux-3/edit-image
1413
+ - https://fal.ai/models/blackforestlabs/flux-3/edit-image/api
package/README.md CHANGED
@@ -5,7 +5,7 @@ provider bindings, and published prices, with the labs that make them and their
5
5
  logos. Every fact cites a first-party source.
6
6
 
7
7
  This package is the data, published unchanged from release
8
- [`v1.20261105.0`](https://github.com/gridaco/models/releases/tag/v1.20261105.0) of [gridaco/models](https://github.com/gridaco/models). It contains no code.
8
+ [`v1.20261105.1`](https://github.com/gridaco/models/releases/tag/v1.20261105.1) of [gridaco/models](https://github.com/gridaco/models). It contains no code.
9
9
 
10
10
  | file | contents |
11
11
  | --- | --- |
@@ -25,7 +25,7 @@ From a CDN, without installing:
25
25
 
26
26
  ```
27
27
  https://cdn.jsdelivr.net/npm/@grida/ai-models@1/models.json
28
- https://cdn.jsdelivr.net/npm/@grida/ai-models@1.20261105.0/models.json
28
+ https://cdn.jsdelivr.net/npm/@grida/ai-models@1.20261105.1/models.json
29
29
  ```
30
30
 
31
31
  ## Versions
package/manifest.json CHANGED
@@ -1,19 +1,19 @@
1
1
  {
2
2
  "name": "grida-models",
3
- "version": "v1.20261105.0",
3
+ "version": "v1.20261105.1",
4
4
  "schema_version": 1,
5
- "commit": "dc130479d081b600afc3ee0357139e0ea7187b87",
6
- "created": "2026-11-05T23:01:53Z",
7
- "source": "https://github.com/gridaco/models/tree/dc130479d081b600afc3ee0357139e0ea7187b87/catalogue",
8
- "digest": "sha256:66c837a0269166699b86b8ef866c781732702a893ff55aa2d80170067ebe2c10",
5
+ "commit": "e6192d001a24cddd30690e1f526b6ed61713b342",
6
+ "created": "2026-11-05T23:05:57Z",
7
+ "source": "https://github.com/gridaco/models/tree/e6192d001a24cddd30690e1f526b6ed61713b342/catalogue",
8
+ "digest": "sha256:2989da18444c3553f97e7631e0e29ff9cc2c0d9a728dd415ded67b60664f5ab7",
9
9
  "counts": {
10
- "cards": 73,
10
+ "cards": 78,
11
11
  "labs": 18
12
12
  },
13
13
  "files": {
14
14
  "PROVENANCE.md": {
15
- "sha256": "97c6de36b9cf3644732b1287375209bb61b9f88e6d05cc5d9611493ad3794d75",
16
- "bytes": 62501
15
+ "sha256": "53822e18fc2123b545e3a39416660ff8e9b79ac3673a8bc71265dd7d2b5e4f4b",
16
+ "bytes": 69757
17
17
  },
18
18
  "README.md": {
19
19
  "sha256": "196afd4ca14e909b5a3adc6bb7bc78f823c45f29163b22047dbc5c30e6448998",
@@ -92,8 +92,8 @@
92
92
  "bytes": 330
93
93
  },
94
94
  "models.json": {
95
- "sha256": "9ab487ff6d9623d439449d192a9ef2af5f05336ec3a056d354d547892e973c6c",
96
- "bytes": 692753
95
+ "sha256": "76bd3bd5ebd97751538237f4a9c0acdecf0c9bf75eb666036eba3e002d62be56",
96
+ "bytes": 760785
97
97
  },
98
98
  "schemas/labs.schema.json": {
99
99
  "sha256": "9ddaedd56826bebdbe72b84f76cfc86df0b5c5902ba9f0259c2ea2cbb9327f99",