[7/8] feat: the remaining upstream model kinds by cunninghamcard-bit · Pull Request #255 · arcships/aimux

cunninghamcard-bit · 2026-10-12T10:47:51Z

Requested by li

Summary

Before: several model kinds that the upstream packages offer are missing: xAI speech, transcription and video; Google and Mistral speech and transcription; Groq transcription; and customProvider. Decisions use aimux's own EvaluationModel next to a DecisionModel with a capability struct, a list of questions with ids, and extra fields beside the call options; decision recordings store a capability snapshot. Realtime transcription streaming is a required trait method whose default returns Unsupported, speech results have no usage, and custom_provider does not forward discovery.

After: the model kinds are ported from the upstream packages, with their upstream tests, and respect the provider boundary rules. Decision is one DecisionModel shaped as @ai-sdk/provider 4.0.22's DecisionModelV4, and decide follows the AI SDK's decide step by step: input validation, unsupported question types rejected before any I/O, two retries by default, then one answer validation. aimux's extensions (images, OpenAI typed values, runtime protocol; confidence, abstention, latency) live in provider options and provider metadata, provider limits are checked before any request, and recordings carry no capability snapshot. Realtime transcription streaming is an optional method, speech results carry the provider's usage, and custom_provider forwards discovery to its fallback.

Type of change

Checklist

cargo fmt --all -- --check passes
cargo clippy --workspace --all-targets -- -D warnings passes
cargo test --workspace passes (cassette playback — no network/keys)
Added or updated cassette tests / contract fixtures for behavior changes
Updated relevant docs (README.md, docs/, rfc/, bindings/README.md)
Commit messages follow Conventional Commits

Notes for reviewers

Position: base PR 6 ([6/8] fix: vendors follow the upstream packages #254); next is PR 8 ([8/8] refactor!: public JSON equals the AI SDK JSON #256).
Supersedes: [v4/p7-model-kinds] feat(providers): the remaining upstream model kinds #241, which GitHub closed when its branch was renamed to match the title.
Replaces: part of Draft fix(providers)!: align vendor packages with upstream and add missing model families #220.
Aligns to: @ai-sdk/provider 4.0.22 decision-model/v4, transcription-model-v4, speech-model-v4-result; ai 7.0.128 decide/; ai 7.0.127 registry/custom-provider.ts, registry/provider-registry.ts.
Closes: V4 alignment: Decision model vs upstream DecisionModelV4 #233 C29, N4, N5, N6, N7, N8, N9; V4 alignment: provider factory, HTTP and auth #230 C21; V4 alignment: vendor package alignment #231 C50, U26 (usage; the tracing half comes with the telemetry split after the migration).
Fork fixes included: none.
Not in this PR:
- decide telemetry (onStart, onEnd, the tracing span): aimux has no telemetry for operations other than language-model calls yet. Registered as a known gap (§9.1).
- The ai/<VERSION> user-agent suffix: no aimux operation adds it, so it is a difference for all operations, not one for decisions.
- The deprecated evaluationModel / doEvaluate aliases and the registry's ?? evaluationModel fallback are not ported, so there is no transitional API.
Behaviour changes (breaking):
- Provider::evaluation_model and EvaluationModel are removed. Provider::decision_model returns Option<Result<Arc<dyn DecisionModel>, _>>, and the registry and custom_provider report a missing one as NoSuchModel { model_type: "decisionModel" }.
- decide(model, DecideOptions) -> DecideResult. Questions are a map keyed by id; answers are {choice, probabilities?}, {score, probabilities?} and {probability}. An OpenAI refusal or a vLLM skipped question fails the decision; there is no partial result.
- decision_model("openai:<id>") is the native OpenAI Decisions API. The Responses adapter that upstream resolves openai:<id> to is OpenAIProvider::evaluation(id).
- C ABI: aimux_decide takes DecideOptions and returns DecideResult JSON. aimux_decision_capabilities keeps its export and signature and returns the model description. A provider without decisions gives AIMUX_E_NO_SUCH_MODEL.
- Bindings: capabilities() becomes describe() (Node describeDecisionModel, Python describe_decision_model); decide takes the V4 options in every language.
- Decision recordings carry no capability snapshot, and the schema stays 2. Decision recordings written by aimux 0.5 load, and replaying one fails with an error that says why.
- decide runs its model call inside the operation settings, as generate_text does, so a router, MoA or trace layer below it sees the decision's retry count and timeout.
- TranscriptionModel::do_stream is optional, and speech results have usage.
Upstream pins: @ai-sdk/provider moves to 4.0.22, the first DecisionModelV4 release. decide follows ai 7.0.128, while VERSIONS.json still pins ai 7.0.127.
Checks: fmt, clippy and cargo test --workspace pass, as do scripts/check_provider_boundaries.sh, scripts/gen_ts_types.py --check and the pin script tests. The Node (ava), Python (pytest), Go, Java and Kotlin suites pass locally; the Swift and Dart decision tests were updated and run in CI.
Merged RFC / docs: docs/api/decision.md rewritten; docs/README.md; alignment doc §0 scope line, §3.1 Decision paragraph, §3.5 decide callbacks, §5 Decision row, §9.1 decide telemetry gap, appendix note; revision lines on RFC-0037 and RFC-0038; the RFC-0023 revision line for aimux 0.5 decision recordings; CHANGELOG.
Draft RFC: RFC-0039: ops、FFI 与八种绑定契约 #209 (Decision in the handle and op tables), RFC-0041: 传输回放与治理证据 #211 (decision recording), RFC-0042: V4 类型边界与统一生成 #212 (SDK baseline).
Size: release artifacts before and after (CI artifacts, zip size):

CI artifact (zip)	master `a8e3a99`	[6/8] `58572b3`	this PR `48083ff`	change vs [6/8]	change vs master
`ffi-x86_64-unknown-linux-gnu`	3.77 MB	4.99 MB	5.11 MB	+123.2 KiB (+2.53%)	+1314.8 KiB (+35.76%)
`ffi-aarch64-apple-darwin`	3.12 MB	4.13 MB	4.24 MB	+104.1 KiB (+2.58%)	+1095.0 KiB (+35.96%)
`ffi-x86_64-pc-windows-msvc`	3.03 MB	4.04 MB	4.15 MB	+102.3 KiB (+2.59%)	+1097.1 KiB (+37.13%)
`node-x86_64-unknown-linux-gnu`	55.30 MB	65.30 MB	66.56 MB	+1225.4 KiB (+1.92%)	+10995.2 KiB (+20.36%)
`node-aarch64-unknown-linux-gnu`	53.25 MB	62.74 MB	63.95 MB	+1185.6 KiB (+1.94%)	+10447.5 KiB (+20.09%)
`node-aarch64-apple-darwin`	15.06 MB	19.34 MB	19.80 MB	+457.4 KiB (+2.42%)	+4631.0 KiB (+31.48%)
`node-x86_64-apple-darwin`	13.60 MB	17.51 MB	17.99 MB	+474.9 KiB (+2.78%)	+4293.4 KiB (+32.33%)
`node-x86_64-pc-windows-msvc`	10.57 MB	13.97 MB	14.33 MB	+352.1 KiB (+2.58%)	+3671.4 KiB (+35.58%)
`node-aarch64-pc-windows-msvc`	9.87 MB	13.13 MB	13.47 MB	+329.9 KiB (+2.57%)	+3514.4 KiB (+36.45%)

…upstream packages - xAI speech, transcription and video models. - Google and Mistral speech and transcription models. - Groq transcription model. - customProvider, ported from upstream. - New models respect the provider boundary rules. OpenAI Chat replay of non-object tool input is pinned to upstream, and the ported Mistral tests name their pinned upstream source. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

…maining-upstream-model-kinds Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Brings [6/8]'s vendor alignment (user-agents from the pinned versions, SigV4 per upstream, Vertex MaaS and xAI, Azure token honours omitted credentials, the restored registry base URL test) together with what [6/8] merged from [5/8] and [1/8]. No conflicts. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

…V4 release @ai-sdk/provider 4.0.22 renamed the experimental EvaluationModelV4 to DecisionModelV4 (src/decision-model/v4/decision-model-v4.ts:6-17) and EvaluationUnsupportedQuestionTypeError to DecisionUnsupportedQuestionTypeError. The decision model follows that release, so the baseline moves to it. - VERSIONS.json pins 4.0.22 with its integrity. - The recorder's package.json and package-lock.json name the same version (regenerated with npm install --package-lock-only --ignore-scripts; the pinned vendor packages keep their own 4.0.21 copies). - aisdk_provider_pin_test fails when the pin drifts from 4.0.22. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

TranscriptionModelV4.doStream is optional (@ai-sdk/provider@4.0.22 src/transcription-model/v4/transcription-model-v4.ts:44-46), and experimental_streamTranscribe rejects a model without it before any I/O (ai@7.0.127 src/transcribe/stream-transcribe.ts:125-138). aimux had a default method that failed with UnsupportedFunctionality, so a missing method could only be found by calling it. - TranscriptionModel::do_stream returns Option<BoxFuture<..>>, None by default, as the alignment doc (§3.1, D22) specifies. - stream_transcribe reports None as UnsupportedFunctionality with the upstream message. - OpenAI, Azure, Google, Vertex, xAI and ElevenLabs return Some; per-model checks (a model ID that cannot stream) stay inside the future, as in the upstream doStream bodies. Groq, Mistral and the other batch-only models return None. - Tests: transcription_model::tests::a_model_without_do_stream_cannot_stream, groq_transcription_test::the_model_has_no_streaming_method. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

A custom provider lost the fallback provider's model discovery, so a registry entry built with custom_provider could no longer list models. Discovery is an aimux extension; it now follows upstream's files/skills forwarding (ai@7.0.127 src/registry/custom-provider.ts:268-283): the configured CustomProviderOptions::discovery, else the fallback's. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

SpeechResult gains usage: Option<JsonObject>, as SpeechModelV4Result's usage?: JSONObject (@ai-sdk/provider@4.0.22 src/speech-model/v4/speech-model-v4-result.ts:23-27). Google speech passes its usageMetadata through unchanged and leaves usage unset when the response has none or null (@ai-sdk/google@4.0.87 src/google-speech-model.ts:356-358, google-speech-api.ts:35). The other vendors report none. Telemetry of speech usage is not part of this change. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Supported question types are a property of the model, as upstream DecisionModelV4.supportedQuestionTypes (@ai-sdk/provider@4.0.22 src/decision-model/v4/decision-model-v4.ts:11), so the recording input no longer stores input.decision_capabilities and replay no longer reads it. The live recording fixture drops the field; the replay mock accepts every question type and size and takes rounding and the probability source from the recorded result. Recording schema stays 3. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

…ader rules Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

aimux's Decision API had its own shape: a question list with ids, Choice options and Score levels, a capability struct, refusals and skips as answers, and extension fields on the call and the result. The AI SDK now has DecisionModelV4 (@ai-sdk/provider@4.0.22 src/decision-model/v4/) and decide (ai@7.0.128 src/decide/decide.ts). This commit makes aimux's Decision that contract, in one step, with no transitional code (D31). It folds N5, N6, N7, C29 and N8 of layer 7. Model and operation (N5, C29): - One DecisionModel per decision-model-v4.ts:6-16: specification_version ("v4"), provider, model_id, supported_question_types and do_decide(&DecisionCallOptions). Call options, questions, answers and the result follow decision-model-v4-call-options.ts:7-14, decision-model-v4-question.ts and decision-model-v4-result.ts:7-56. - decide follows decide.ts:76-185: validate the input (validate-decision.ts:50-125), reject unsupported question types before any I/O with the DecisionUnsupportedQuestionTypeError message (decision-unsupported-question-type-error.ts:18-31), retry (default 2), validate the answers once after the retries (validate-decision.ts:176-298, tolerance 1e-6 and rounding decimals 0..15), complete total_tokens only when both counts are known and default the response timestamp and model. Provider lookup (N7): - Provider::evaluation_model and the EvaluationModel trait are removed. Provider::decision_model returns Option<Result<Arc<dyn DecisionModel>>>, None by default. The registry and custom_provider have one decision lookup and report None as NoSuchModel("decisionModel") (provider-registry.ts:447-461, custom-provider.ts:241-258). The deprecated evaluationModel / doEvaluate aliases are not ported. - EvaluationLanguageModel is DecisionLanguageModel, with the default provider "{language provider}.decision" and the response format name "decision" (provider-utils 5.0.55 decision-language-model.ts:24, :158). Google's decision_model is that adapter ("google.evaluation", google 4.0.87 google-provider.ts:438-441); OpenAI's stays the native Decisions API, and OpenAIProvider::evaluation builds the Responses adapter ("openai.evaluation", openai 4.0.83 openai-provider.ts:355-358). API (N6): - No capability struct. Limits are checked in do_decide before HTTP. - Request extensions move to provider_options.<provider>: images, OpenAI choice_values (typed native values) and safety_identifier, runtime protocol. Result extensions move to provider_metadata.<provider>: probability_source, per-answer confidence, OpenAI typed values, Laya abstentions, latency_ms. - An OpenAI refusal or a vLLM skipped question fails the decision; a V4 decision answers every question. - Recordings hold the redacted DecideOptions and the DecideResult, with no capability snapshot; mock replay returns the recorded result. - FFI: aimux_decide takes DecideOptions and returns DecideResult; aimux_decision_capabilities keeps its export and returns the model's V4 properties. Bindings: capabilities() is describe() (Node describeDecisionModel, Python describe_decision_model). Fixtures (N8): the decision fixtures and the live Jev recordings were converted to the V4 shape by a deterministic script. Re-decoding both live Jev wire responses through the new adapter gives the converted results (decision_recording_test). Tests pinned to upstream: decision_test (questions_have_the_v4_shape, invalid_input_is_rejected_before_the_model_is_called, unsupported_question_types_fail_before_the_model_is_called, transient_failures_are_retried_twice_by_default, answers_are_validated_once_after_the_retries, the_result_completes_usage_and_response, invalid_answers_report_the_upstream_messages, ...); custom_provider_test (decision_model::*, decision_*, registry_reports_missing_decision_models); decision_language_model_test (provider_defaults_to_the_language_provider_with_a_decision_suffix, errors_use_the_upstream_decision_messages, the_language_model_call_joins_the_decision_recording); provider tests for OpenAI, Google, Jev and the runtimes; FFI, Node and Python tests. Docs: docs/api/decision.md describes the V4 API; the alignment doc §3.1, §3.5, §5 and §9.1 (decide telemetry is a known gap); revision lines on RFC-0037 and RFC-0038; CHANGELOG Breaking entries. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

…settings, factories) fx3/p6 at 3f389ba: layer 6 after its rounds 3 and 4, and the merge of layer 5, which carries layers 2 to 4. Conflicts resolved: - provider-utils lib.rs: layer 7's decision_language_model module and layer 6's download_blob; the download_guard re-export goes, as in layer 5. - decision_support.rs: layer 7's provider_metadata and response_info, built on layer 3's IndexMap SharedProviderMetadata. - jev.rs: layer 6's jev_failed_response_handler and layer 7's rounding doc. - openai/mod.rs imports: layer 7's DecisionLanguageModel, without the load_optional_setting import layer 5 dropped. - Recorder pins: package.json pins @ai-sdk/provider 4.0.22 (layer 7) and provider-utils 5.0.55 (layer 2). package-lock.json was regenerated from layer 6's lock with npm install --package-lock-only --ignore-scripts: the only change is the top-level provider moving to 4.0.22, with each vendor keeping its own nested copy. VERSIONS.json merged cleanly (18 entries). - CHANGELOG: both sides' Added bullets; layer 6's Fixed section. - Generated TS types: layer 6's, regenerated in a follow-up commit. Follow-up commits fix the code that does not compile against the merged layers. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

The merge of [6/8] left layer 7's decision and transcription code on the pre-merge shapes: - indexmap is a workspace dependency since layer 3; the duplicate key in aimux-core is gone and the other crates take the workspace entry. - WrapProvider forwards decision_model with the provider factory's Option<Result<..>> signature. - CallOptions no longer carries call_id (layer 4): the decision language model's exchanges share the decision's recording context, which supplies the call id. - Provider metadata and options are IndexMap namespaces (layer 3). - The google, groq, mistral and xai transcription models refuse AudioInput::Url (layer 5) with AudioInput::url_error(), as every transcription model does; only transcribe downloads a URL. - Tests: SharedProviderOptions literals, download: None in TranscriptionCallOptions, do_stream's Option (C21), and the replay tool's decision_model probe; the two test modules in transcription_model.rs are one. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

google, groq, mistral and xai each get url_audio_is_refused_before_any_request: an AudioInput::Url fails with the url error before any request is sent, as layer 5 pins for the transcription models it already had. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Layer 4 moved max_retries and timeout out of CallOptions into the operation's OperationSettings, which generate_text and stream_text scope around their model calls. decide now scopes its model call the same way, so a language model under DecisionLanguageModel (a router or MoA reading the child retry count, a TraceLayer) sees the decision's settings. A direct do_decide call has none, as a direct do_generate. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

… order V4 decision questions and choice criteria are maps, and the OpenAI request lists them in the caller's order. The test re-serialized the fixture's options through org.json's JSONObject, which does not keep key order, so the request listed score before choice and failed the comparison with the fixture. It now passes the options' raw text. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Layer 4 rewrote the Rust doc of applyToolCallRepair (it repairs the call in the result's steps) without regenerating index.d.ts, so every napi build rewrote the file. This is the napi build's output for the merged source; only that doc comment changes. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

fx3/p6 at 8bef2c9, which carries fx3/p4 at 4ce9aa0 and fx3/p3 at 35711cb: - the Node reasoning_replay test sends its tool call with the tool result (layer 3's prompt check); - the Node and Python DeepSeek tests read the provider's raw usage from steps[0].usage.raw, since the summed usage carries none (layer 4); - bindings/node/index.d.ts gets layer 4's regeneration, the same hunk as 6aa327d, so it merges without a conflict. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

…order The same order loss as the Kotlin test before efd10d6: the test re-encoded the fixture's options with JSONSerialization, whose dictionaries keep no key order, and compared the request as an NSDictionary, whose arrays are ordered. V4 questions and choice criteria are maps, and the OpenAI request lists them in the caller's order. The test now passes the raw text of the fixture's options object. Not compiled here (no Swift toolchain); the brace matcher was checked against the fixture with an equivalent script. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

fx3/p6 at a382242, which merges fx3/p5 at 9d01eb5 (recording schema 2; recordings written by master load and fail at replay time, naming aimux 0.5 or earlier, where this version cannot replay them). Conflicts, resolved to layer 7's decision shape (N9: no capability snapshot; DecisionModelV4) with layer 5's schema 2: - aimux-core/src/decision_replay.rs: layer 7's MockDecisionReplayModel (no capabilities). A master decision recording (no provider_id) loads; replay_decision_with_model reports its pre-V4 options as written by aimux 0.5 or earlier, and the mock's no-match error counts such recordings. - aimux-providers/src/replay.rs: layer 7's Arc return and NoSuchModel, with ProviderRecord::registry_id. - jev_systemone_live.jsonl: layer 7's converted content at schema 2; the .md note joins both texts. - docs/api/decision.md: layer 7's recording text at schema 2, plus how aimux 0.5 decision recordings behave. Also: master_recording_test's decision test follows layer 7 (the replay error instead of the capability error); CHANGELOG ("schema stays 2", pre-V4 decision options and results) and the RFC-0023 revision line no longer mention input.decision_capabilities or provider.profile. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

The provider-factory bullet still said decision recordings keep the capabilities under input; since decisions follow DecisionModelV4 they carry none, as the DecisionModelV4 entry says. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

rustdoc -D warnings rejects intra-doc links to private items. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

aimux_stream_text runs the whole stream future on the caller's thread (ffi_block_on). With the test-only override at opt-level 0, one stream through the C ABI needs about 530 KB of stack on the AI SDK branches (about 300 KB on master), more than the 512 KB stack of a Swift cooperative thread: the async Swift tests died with SIGBUS on #255 and #259 once they carried this workflow. At opt-level 1 the same stream needs about 110 KB; the shipping profile needs about 160 KB. The library is also 40% smaller, and the build is still far from the shipping profile's fat LTO with one codegen unit. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0112xDERMUyD5JALkFhnsdWS

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

cunninghamcard-bit · 2026-10-14T05:52:46Z

补充 #233 N8 在本 PR（22db0867）中的实际验证方式，状态为已实现、待合并：沿用既有真实 HTTP 录制，将 options/result 离线、确定性转换为 DecisionModelV4 新形状；原始 HTTP 请求/响应保持不变，录制仅保留 provider identity，不含能力快照。详见 fixture 说明。

recorded_native_calls_replay_offline_and_rebuild_uses_registered_endpoint 覆盖录制结果离线回放，以及在本地 HTTP 服务回放既有 wire、从 registry 重建 provider 后比较新形状结果（比较时排除 timestamp/headers）；另有录制生命周期和重试回放覆盖。因此 #233 原「重新录制」措辞应按此对齐，不要求新增网络录制，也不把未重新录制作为 bug。以上仅据远程源码核对，本次未运行测试。

cunninghamcard-bit mentioned this pull request Oct 12, 2026

[1/8] chore: CI on every PR and test fixture cleanup #249

Open

6 of 12 tasks

cunninghamcard-bit changed the title ~~[V4 7/8] feat: the remaining upstream model kinds~~ feat: the remaining upstream model kinds Oct 12, 2026

cunninghamcard-bit changed the title ~~feat: the remaining upstream model kinds~~ [7/8] feat: the remaining upstream model kinds Oct 12, 2026

fix: carry provider stack repairs into model kinds

c8d773a

cunninghamcard-bit mentioned this pull request Oct 12, 2026

[8/8] refactor!: public JSON equals the AI SDK JSON #256

Open

9 of 12 tasks

claude added 22 commits October 12, 2026 13:02

Merge branch 'v4/6-vendors-follow-the-upstream-packages' into v4/7-re…

5438b98

…maining-upstream-model-kinds Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Merge [6/8] into [7/8]: per-package user-agents and upstream error/he…

37255a7

…ader rules Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Merge [6/8] into [7/8]: CHANGELOG wording for legacy namespace keys

9fcc401

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

Merge [6/8] into [7/8]: rustdoc link fixes

ee60e4e

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

docs: plain code spans for the private stream_transcription methods

48083ff

rustdoc -D warnings rejects intra-doc links to private items. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

cunninghamcard-bit mentioned this pull request Oct 14, 2026

[6/8] fix: vendors follow the upstream packages #254

Open

10 of 12 tasks

cunninghamcard-bit marked this pull request as ready for review October 14, 2026 03:06

Merge [6/8] into [7/8]: the CI speed-up (#262)

f80fff2

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

cunninghamcard-bit mentioned this pull request Oct 14, 2026

ci: speed up test builds, deduplicate caches, and use prebuilt cargo-ndk #262

Open

Merge [6/8] into [7/8]: the CI fix (#262)

22db086

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01A3WxDBWBA4ituJjXsHaEAa

This was referenced Oct 14, 2026

V4 alignment: vendor package alignment #231

Open

V4 alignment: provider factory, HTTP and auth #230

Open

V4 alignment: Decision model vs upstream DecisionModelV4 #233

Open

V4 alignment: public JSON, tools and lost tests #232

Open

Navigation Menu

Repository navigation

[7/8] feat: the remaining upstream model kinds - #255

[7/8] feat: the remaining upstream model kinds#255
cunninghamcard-bit wants to merge 26 commits into
v4/6-vendors-follow-the-upstream-packagesfrom
v4/7-remaining-upstream-model-kinds

cunninghamcard-bit commented Oct 12, 2026 •

edited

Loading

Uh oh!

cunninghamcard-bit commented Oct 14, 2026

Uh oh!

This branch has not been deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

cunninghamcard-bit commented Oct 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Type of change

Checklist

Notes for reviewers

Uh oh!

cunninghamcard-bit commented Oct 14, 2026

Uh oh!

This branch has not been deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

cunninghamcard-bit commented Oct 12, 2026 •

edited

Loading