Commit Graph

29 Commits

Author SHA1 Message Date
Armin Ronacher 2e4ad6a094 fix(ai): clamp OpenAI Responses max output token floor
closes #6265
2026-07-06 02:21:15 +02:00
Vegard Stikbakke 23d1462611 fix(ai): rotate stale Codex websocket sessions
closes #6268
2026-07-03 15:30:41 +02:00
Vegard Stikbakke 114bacf349 fix(ai): enable Bedrock prompt caching for Claude 5
closes #6235
2026-07-02 11:00:20 +02:00
Vegard Stikbakke 0ac3cfe09b feat(ai): use zstd compression for codex sse transport 2026-07-01 08:42:46 +02:00
Vegard Stikbakke a3cc169d97 fix(ai): avoid codex user-agent race
The Codex provider previously loaded node:os asynchronously during module evaluation. A fresh process that imports the provider and immediately starts an SSE request can build headers before that promise callback runs, so the first request reports User-Agent: pi (browser) in Node/Bun.

Reproduced against parent fd6659dd with a stubbed-fetch harness: 50/50 immediate first requests reported pi (browser) in both Node and Bun. The same harness on this fix reports the OS-specific user agent 50/50 in both runtimes.

Use process.getBuiltinModule("node:os") behind the existing Node/Bun runtime guard so OS metadata is available synchronously while still avoiding top-level runtime Node builtin imports that break browser/Vite builds.
2026-07-01 08:41:38 +02:00
Mario Zechner 5c1a2977bc fix(ai): update generated model catalogue 2026-06-30 22:12:48 +02:00
Mario Zechner 6fbeba51af Merge pull request #5832 from stephanmck/fix/provider-error-body-passthrough-5763
fix(ai): surface provider HTTP error body instead of opaque SDK message
2026-06-29 16:58:24 +02:00
Vegard Stikbakke b91bdd5a3e fix(ai): preserve Z.AI thinking content
closes #6083
2026-06-29 08:40:20 +02:00
Armin Ronacher 54113731b2 fix(ai): use HTTP timeout for Codex SSE headers
Refs #4945
2026-06-28 19:57:47 +02:00
Mario Zechner 09f1059575 fix(ai): clamp streamSimple max tokens
Clamps streamSimple max-token defaults against estimated context, addressing #5595.

closes #6061
2026-06-25 14:10:28 +02:00
Mario Zechner f78b163713 fix(ai): revert minimax max token clamp 2026-06-25 13:36:04 +02:00
Stephan Schneider 62fad94f1a fix(ai): surface provider HTTP error body instead of opaque SDK message
Add a shared normalizeProviderError helper in packages/ai/src/utils/error-body.ts
and route the 8 body-blind / status-only providers through it (amazon-bedrock,
azure-openai-responses, google, google-vertex, images/openrouter,
openai-codex-responses, openai-completions, openai-responses). Non-schema 4xx/5xx
responses from proxies / gateways now show the real reason carried in the response
body alongside the HTTP status, instead of "403 status code (no body)" or
"Unknown: UnknownError".

The helper probes status (statusCode, status, $metadata.httpStatusCode,
$response.statusCode) and body (body, parsed error object, $response.body) across
the Mistral, OpenAI, Google, and Bedrock SDK shapes, truncates the body at a 4000
char cap, and preserves error.message when the SDK already folded the body in
(Anthropic / Google happy path). mistral.ts and anthropic.ts are left untouched.
Provider prefixes and the OpenRouter metadata.raw append are preserved.

closes #5763
2026-06-25 12:07:51 +02:00
Mario Zechner b940c52e7e fix(ai): clamp MiniMax shared budget max tokens
closes #6061
2026-06-25 12:00:37 +02:00
Mario Zechner 8c9dbffa36 fix(ai): preserve responses reasoning for out-of-order items
closes #6009
2026-06-25 11:57:52 +02:00
Mario Zechner d7868b0998 feat(ai): add reasoning token counts to Usage
Add optional reasoning?: number to Usage as a subset of output. Populate
for Anthropic (output_tokens_details.thinking_tokens), OpenAI
Responses/Codex/Azure (output_tokens_details.reasoning_tokens), OpenAI
Completions (completion_tokens_details.reasoning_tokens), and Google
Generative AI / Vertex (thoughtsTokenCount). Bedrock Converse and Mistral
do not return a reasoning breakdown, so they stay unset.

closes #6057
2026-06-25 10:35:02 +02:00
Mario Zechner a2e3e9d8b2 Merge pull request #6004 from gukoff/support-azure-foundry-endpoints
feat: Normalize modern Microsoft Foundry Responses API endpoints
2026-06-24 10:38:57 +02:00
Mario Zechner e1a2dc04f4 fix(ai): restore detectCompat runtime fallback in openai-completions
refs #6020
2026-06-23 23:29:44 +02:00
Mario Zechner 6184307c39 fix(ai): require explicit anthropic compat metadata 2026-06-23 21:04:59 +02:00
Armin Ronacher 828493b37c fix(ai): unblock release provider tests 2026-06-23 19:57:29 +02:00
Mario Zechner cd95c2749f fix(ai): require OpenAI Responses terminal events 2026-06-23 16:35:45 +02:00
Konstantin Gukov e3dcb244f8 Support modern Microsoft Foundry Responses API endpoints:
- support *.ai.azure.com endpoints
- normalize away the /responses postfix in the query path which by
  default is added on Microsoft Foundry UI
2026-06-23 15:38:43 +02:00
Mario Zechner 129eb460cd feat(ai): complete models runtime migration 2026-06-23 15:29:46 +02:00
Mario Zechner 6a4813a7ca Merge remote-tracking branch 'origin/main'
# Conflicts:
#	packages/ai/CHANGELOG.md
2026-06-23 13:52:09 +02:00
Mario Zechner 5a8ea0bc81 fix(ai): honor scoped AWS profile in Bedrock endpoint resolution 2026-06-23 13:36:44 +02:00
Armin Ronacher d0e0b84cb9 fix(ai): reconnect Codex websocket on connection limit
closes #5973
2026-06-23 00:17:37 +02:00
Mario Zechner abbd911693 Merge main into model-registry 2026-06-22 14:00:18 +02:00
Mario Zechner 827fe1e2c4 feat(ai): ImagesModels collections mirroring the chat-side design
createImagesModels()/ImagesProvider/createImagesProvider() give image
generation the same shape as chat: sync model reads, explicit async
refresh(provider?) with in-flight dedupe, provider-resolved auth, and
never-rejecting generateImages() (failures return AssistantImages with
stopReason error). Auth resolution is shared with the chat side via the
free-standing resolveProviderAuth() in auth/resolve.ts, which also owns
ModelsError; both collections pass their store/context as arguments.

The OpenRouter implementation moves to api/openrouter-images.ts with a
lazy wrapper; openrouterImagesProvider() factory plus
builtinImagesProviders()/builtinImagesModels() land in providers/all.
The ImagesProvider id type alias is renamed to ImagesProviderId
(mirror of Provider -> ProviderId). The old global image API
(getImageModel*, generateImages, registerImagesApiProvider) stays on
/compat, its registration shim repointed at the moved implementation.

README: Quick Start uses builtinModels(), the full streaming event
switch, image generation and the development checklist are restored in
full, image generation documents the new collections with compat noted
for the old API, plus the review fixes (builtinModels options,
credential-store mention for browsers, ImagesModels notes).
2026-06-11 00:27:43 +02:00
Mario Zechner ba93da9a93 feat(ai): move API implementations to src/api with lazy wrappers (phase 2)
Stream implementations move from src/providers/ to src/api/, renamed by
API id (anthropic.ts -> anthropic-messages.ts, google.ts ->
google-generative-ai.ts, mistral.ts -> mistral-conversations.ts,
amazon-bedrock.ts -> bedrock-converse-stream.ts). Every module now
exports exactly stream/streamSimple; shared helpers move alongside.

New ProviderStreams dispatch contract in types.ts, lazyApi() wrapper in
api/lazy.ts, and one .lazy.ts wrapper per API. Bedrock's wrapper keeps
the node-only variable-specifier import and setBedrockProviderModule()
(now taking ProviderStreams).

providers/register-builtins.ts deleted; interim until the compat
entrypoint lands, builtin api-registry registration lives in stream.ts
and lazy wrappers are exported from the root barrel. Old per-API lazy
exports (streamAnthropic, ...) are gone; package.json subpaths retarget
to dist/api/.
2026-06-10 20:08:59 +02:00
Mario Zechner f63095cfff feat(ai): add Models runtime with provider-owned auth (phase 1)
New Models/MutableModels/createModels collection: provider map, async
model listing (best-effort aggregation), getAuth decision tree with
double-checked locked OAuth refresh, stream/complete with per-field
auth merge over lazyStream.

Auth substrate: ProviderAuth { apiKey?, oauth? }, one type-tagged
credential per provider, CredentialStore (read/modify/delete; modify
is the only write path, serialized RMW), OAuthAuth login/refresh/toAuth
split, prompt()/notify() login callbacks, browser-safe default
AuthContext.

types.ts: Provider alias renamed to ProviderId; ApiOptionsMap and
ApiStreamOptions<TApi> for typed per-API stream options; hasApi()
runtime narrowing guard.
2026-06-10 18:49:10 +02:00