agent/pi - pi - Penguin

feat(ai): provider factories, per-provider catalogs, createProvider (phase 3)

Auth helpers in src/auth/helpers.ts: envApiKeyAuth() (stored key wins,
then env vars in order, with secret-prompt login) and lazyOAuth()
(flow loads on first use through bundler-opaque dynamic imports in
utils/oauth/load.ts; the OAuthAuth flow exports land in phase 4).
There is no OAuth factory toggle: providers that support OAuth always
attach it, advertising costs nothing until login/refresh runs.

createProvider() in models.ts builds providers from parts: single API
implementation or a map dispatched on model.api (mixed-API providers
like opencode and github-copilot); unknown api yields a stream error.

generate-models.ts now emits one providers/<id>.models.ts catalog per
provider (35 files, biome-excluded like models.generated.ts) and
models.generated.ts becomes a generated aggregator, so importing one
provider factory pulls one catalog. Typed getModel globals unchanged.

One factory per built-in provider under src/providers/: envApiKeyAuth
for standard providers, OAuth for anthropic/openai-codex/github-copilot,
ambient ApiKeyAuth for amazon-bedrock (AWS env/profile/IAM) and
google-vertex (explicit key or ADC+project+location).

providers/all.ts: builtinProviders(), builtinModels(), getBuiltin*
re-exports. fauxProvider() factory returns a real Provider for tests;
legacy registerFauxProvider() unchanged.

Mario Zechner · 2026-06-10 20:33:20 +02:00

fec0c3d12f

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/.

Mario Zechner · 2026-06-10 20:08:59 +02:00

ba93da9a93

fix(ai): omit disabled thinking for Claude Fable 5

Armin Ronacher · 2026-06-10 00:27:11 +02:00

9ccfcd7cfc

fix(ai): correct Azure gpt-5.4/5.5 context window and gpt-5-pro maxTokens

Azure Foundry deploys gpt-5.4 and gpt-5.5 with a 1,050,000 context
window, but the Azure provider cloned OpenAI-direct's 272k API limit.
Override the context window in the Azure derivation.

Also fix gpt-5-pro maxTokens, which upstream metadata set to 272000 (a
duplicate of the input sub-limit) instead of the actual 128000 max
output; corrected at the source so OpenAI-direct and Azure both match.

closes #5559

Armin Ronacher · 2026-06-09 22:57:20 +02:00

6b5923f107

feat(ai): add Claude Fable 5 metadata

Armin Ronacher · 2026-06-09 22:42:48 +02:00

5a9d72ea02

fix(ai): map OpenCode max tokens

closes #5331

Mario Zechner · 2026-06-09 13:41:08 +02:00

3d02d1da11

fix(ai): disable Moonshot thinking when requested

closes #5531

Mario Zechner · 2026-06-09 12:23:41 +02:00

2326d5cb4a

Update generate-models.ts

Removed together.ai MiniMaxAI/MiniMax-M2.5 due to it not being supported for serverless inference.

 Error: 400 Unable to access non-serverless model MiniMaxAI/MiniMax-M2.5. Please visit
 https://api.together.ai/models/MiniMaxAI/MiniMax-M2.5 to create and start a new dedicated endpoint for the model.

Eric Henry · 2026-06-08 08:32:16 -05:00

2527441b40

Merge pull request #5333 from vastxie/zai-coding-cn-provider

feat(ai): add ZAI Coding Plan China provider

Mario Zechner · 2026-06-03 01:07:01 +02:00

564ad70fb8

fix(ai): remove stale codex models

Mattia Cerutti · 2026-06-03 00:09:30 +02:00

83afcdc24f

feat(ai): add ZAI Coding Plan China provider

弎水 · 2026-06-02 23:39:56 +08:00

51df39b9b9

Add Ant Ling provider

Mario Zechner · 2026-06-02 16:37:44 +02:00

25a4a8ed1e

Add MiniMax-M3 to direct minimax providers

closes #5313

Mario Zechner · 2026-06-02 16:00:29 +02:00

7e72ca47c8

Fix OpenRouter Kimi reasoning compat

closes #5309

Mario Zechner · 2026-06-02 15:36:52 +02:00

2125221bfc

Restore NVIDIA Qwen 3.5 122B model

Mario Zechner · 2026-06-02 15:10:36 +02:00

6e269edfd0

Add NVIDIA NIM provider

Mario Zechner · 2026-06-02 15:09:36 +02:00

6014801229

Fix OpenAI GPT-5.5 thinking metadata

closes #5243

Mario Zechner · 2026-06-01 00:39:48 +02:00

f429ddbffd

fix(ai): suppress deprecated temperature param for Claude Opus 4.7+

Michael Yu · 2026-05-31 16:20:22 +08:00

40832d1957

Fix OpenRouter Kimi K2.6 developer role

closes #5159

Mario Zechner · 2026-05-29 23:41:14 +02:00

a213abb97e

fix(ai): handle OpenCode Kimi reasoning params

Mario Zechner · 2026-05-29 23:27:48 +02:00

ba2d313858

fix(ai): handle OpenCode reasoning params

Armin Ronacher · 2026-05-29 12:21:37 +02:00

4faac05419

Fix OpenRouter DeepSeek V4 xhigh reasoning

closes #4801

Mario Zechner · 2026-05-28 23:56:59 +02:00

0127caea7b

Fix Kimi and Xiaomi model metadata

Closes #5075, closes #5078. Refs #5087.

Mario Zechner · 2026-05-28 22:19:47 +02:00

97ef317389

Update Claude Opus and GPT thinking metadata

Mario Zechner · 2026-05-28 21:51:55 +02:00

bfa3d1fa60

Fix Anthropic empty thinking signature replay

closes #4464

Mario Zechner · 2026-05-28 12:04:33 +02:00

458a7bc27c

fix(ai): correct Codex Spark context window

closes #4969

Armin Ronacher · 2026-05-26 09:08:40 +02:00

71446c6c2b

Support adaptive thinking for Anthropic-compatible aliases

closes #4790

Mario Zechner · 2026-05-22 18:30:51 +02:00

d801d88a11

chore: pin dependencies and use native TypeScript

Mario Zechner · 2026-05-20 12:46:17 +02:00

2e02c74dcb

chore(ts): use source import extensions

Armin Ronacher · 2026-05-20 00:04:03 +02:00

ae9450dc51

fix(ai): add Xiaomi reasoning replay compat

closes #4678

Mario Zechner · 2026-05-18 11:15:20 +02:00

b8f51957a0

fix(ai): switch xiaomi models to openai completions

closes #4505

Mario Zechner · 2026-05-18 00:14:31 +02:00

ed3904ddd3

Remove openai-codex fast model variants, they do not work

Mario Zechner · 2026-05-18 00:02:52 +02:00

266234047a

Merge pull request #4603 from mattiacerutti/fix/openai-codex-model-list

fix(ai): update OpenAI Codex model list

Mario Zechner · 2026-05-17 20:54:53 +02:00

a01cf7afae

fix(ai): map copilot gpt minimal thinking to low

Mattia Cerutti · 2026-05-17 12:18:50 +02:00

485afc9c3f

fix(ai): update OpenAI Codex model list

Mattia Cerutti · 2026-05-17 03:12:19 +02:00

1af823be9d

Release v0.74.1

Mario Zechner · 2026-05-17 01:35:45 +02:00

2c708492e3

fix(ai): mark inception/mercury-2 thinkingLevelMap.off as null

Mercury 2 in instant mode (reasoning_effort: "none") disables tool calling.
The openai-completions provider hardcodes {reasoning:{effort:"none"}} when no
explicit reasoning level is passed and thinkingLevelMap.off isn't null
(openai-completions.ts:575), so every caller that doesn't opt in to a level
silently breaks Mercury 2's agentic use cases.

Setting thinkingLevelMap.off = null on the Mercury 2 catalog entry causes the
provider to omit the reasoning param entirely, letting Mercury 2's own default
take over. Low/medium/high pass through verbatim; OpenRouter normalizes them
to Mercury's vocabulary.

Prefix-matched on "inception/mercury-2" so future Mercury 2 variants on
OpenRouter inherit the fix.

Apoorv Saxena · 2026-05-13 19:17:36 +05:30

e2b69a0bb1

fix(ai): add session affinity and compat fixes for Fireworks provider caching

Fireworks prompt caching is enabled by default (automatic prefix matching),
but on serverless infrastructure, requests hit random replicas. Without
session affinity, the per-replica cache misses, negating cache hit rates
and the discounted cacheRead pricing.

Changes:

- Add sendSessionAffinityHeaders and supportsCacheControlOnTools
  to AnthropicMessagesCompat interface
- Send x-session-affinity header for Fireworks (and Cloudflare AI
  Gateway Anthropic) when sessionId is available and caching is enabled
- Omit cache_control on tool definitions for Fireworks (unsupported
  per https://docs.fireworks.ai/tools-sdks/anthropic-compatibility)
- Default supportsEagerToolInputStreaming to false for Fireworks
  (unsupported field)
- Default supportsLongCacheRetention to false for Fireworks
  (cache_control.ttl not supported)
- Add compat settings to Fireworks models in generate-models.ts
- Update generated models with Fireworks compat settings
- Add integration tests for session affinity and tool compat

Refs: https://docs.fireworks.ai/guides/prompt-caching
Refs: https://docs.fireworks.ai/tools-sdks/anthropic-compatibility

yanirz · 2026-05-10 00:11:36 +02:00

99dc6fcec8

feat(ai): add Together AI provider

Mario Zechner · 2026-05-08 16:44:18 +02:00

7adb8e7634

Merge pull request #3887 from cristinaponcela/feat/image-outputs

feat: image content

Mario Zechner · 2026-05-08 15:57:06 +02:00

9751057be9

fix(ai): disable OpenAI reasoning where supported

Mario Zechner · 2026-05-07 22:52:54 +02:00

783e96a144

Merge remote-tracking branch 'upstream/main' into feat/image-outputs

Cristina Poncela Cubeiro · 2026-05-07 16:46:17 +02:00

ffdf426e33

fix(ai): normalize kimi k2p6 alias to kimi-for-coding closes #4218

Armin Ronacher · 2026-05-06 11:34:49 +02:00

ace3fd7e30

Merge branch 'main' into feat/image-outputs

Cristina Poncela Cubeiro · 2026-05-04 16:42:51 +01:00

5731e13a61

feat: image models

Cristina Poncela Cubeiro · 2026-05-04 17:07:02 +02:00

63c61aac6f

revert

Cristina Poncela Cubeiro · 2026-05-04 15:43:45 +02:00

cbf3c333ef

feat(ai): switch xiaomi default to api billing, add per-region token plan providers (#4112 )

Built-in `xiaomi` provider now targets the API billing endpoint (https://api.xiaomimimo.com/anthropic) — a single stable URL for keys issued at platform.xiaomimimo.com. The Token Plan endpoints are exposed as three sibling providers, each with its own env var:

- xiaomi-token-plan-cn: XIAOMI_TOKEN_PLAN_CN_API_KEY
- xiaomi-token-plan-ams: XIAOMI_TOKEN_PLAN_AMS_API_KEY
- xiaomi-token-plan-sgp: XIAOMI_TOKEN_PLAN_SGP_API_KEY

BREAKING CHANGE: users who previously set XIAOMI_API_KEY against the Token Plan AMS endpoint must move to xiaomi-token-plan-ams and set XIAOMI_TOKEN_PLAN_AMS_API_KEY. This also resolves the 401 reported by on #4005, where a platform.xiaomimimo.com key fails against the Token Plan endpoint.

closes #4082

Jake Jia · 2026-05-03 12:57:11 +02:00

693888ac47

fix(ai): fix mismatch between models.dev and OpenCode Go (Qwen3.5/3.6, MiniMax M2.7) (#4110 )

Jakub Synowiec · 2026-05-03 00:41:55 +02:00

c8edb256b9

fix(ai): use Xiaomi Token Plan Anthropic endpoint

closes #3912

Mario Zechner · 2026-05-02 01:36:34 +02:00

c0e046990e

feat: add model thinking level metadata

closes #3208

Mario Zechner · 2026-05-02 01:21:06 +02:00

80f06d3636

169 Commits