ai_sdk_dart

AI SDK Dart 2.0 is a coordinated major release across all 13 packages.

Highlights

• V4 is now the only language-model provider contract.
• Generation and streaming gain scoped timeouts, provider-backed cancellation, stable runtime context and telemetry APIs, structured output improvements, and safer streamed tool execution.
• Provider usage reports nested input/output token details, including cache reads and writes where available.
• Flutter UI gains polished tool approval, source citation, reasoning, streaming, error, and long-history experiences.
• MCP transports and lifecycle handling are substantially hardened.

Breaking changes

Upgrade every ai_sdk package together. The obsolete V3 language-model contract, experimental API names, video surface, and old provider call fields were removed without compatibility shims.

Read the 2.0 migration guide before upgrading.

Packages

All 13 packages are published on pub.dev.

What's Changed

• fix(openai): surface reasoning/thinking deltas on streaming + non-streaming paths (fixes #3) by @codenameakshay in #7
• chore: add SessionStart hook to provision Flutter via FVM on the web by @codenameakshay in #5
• feat(usage): map provider cache tokens into inputTokenDetails by @codenameakshay in #9
• Prepare the next release with V4 providers and hardened UX by @codenameakshay in #10
• Prepare the coordinated 2.0.0 release by @codenameakshay in #11

Full Changelog: 1.2.0...2.0.0

• Migrated generation, streaming, middleware, mocks, and tools to the V4 language-model contract.
• Added scoped total/step/chunk/tool timeouts, provider-backed cancellation, stable runtimeContext/telemetry names, explicit approval policies, and machine-readable structured-stream benchmarks.
• Tools now execute only after a complete validated stream tool call, not when its input stream ends.

Coordinated 1.2.0 release across all 13 packages — now live on pub.dev. A v6 hardening pass that makes the AI SDK Dart port more complete, honest, and tested.

Highlights

🚨 Typed provider API errors (new)

Non-2xx provider responses now throw a typed AiApiCallError carrying the provider's message, type, code, statusCode, raw body, and an isRetryable flag — consistently across every provider. The AiSdkError hierarchy moved into ai_sdk_provider (re-exported by ai_sdk_dart, so the public surface is unchanged).

🧩 Shared ai_sdk_openai_compatible base (new package)

A shared OpenAI Chat Completions implementation now powers OpenAI / Azure / Groq / Mistral. All 8 providers support tool use + multimodal image input; Cohere & Ollama gained them natively.

🔌 MCP

Real Server-Sent-Events streaming transport (SseClientTransport) plus a plain request/response HttpClientTransport; web-safe — dart:io is isolated behind conditional imports, so the client runs on Flutter web.

📱 Flutter UI hub

19 prebuilt, themeable Material widgets plus hardened controllers (ObjectStreamController.submit, ChatController.isStreaming, streaming-error surfacing). A restrained Material 3 redesign — bubbleless assistant transcript, an AiMotion motion/haptics vocabulary, and a first-class reduced-motion path. Fixed a PressableScale crash-on-dispose under reduced motion.

✅ Correctness

onAbort now interrupts an in-flight streamText; stopWhen can extend the tool loop past maxSteps. Removed the unusable generateVideo / VideoModelV1 (no backing provider). OpenAI image generation no longer sends response_format to gpt-image-1.

🧪 Quality

1,057 tests (924 Dart + 133 Flutter), 99.9% line coverage (11 of 12 packages at 100%), enforced by a CI coverage gate.

Packages (all published at 1.2.0)

ai_sdk_provider · ai_sdk_dart · ai_sdk_openai_compatible · ai_sdk_openai · ai_sdk_anthropic · ai_sdk_google · ai_sdk_azure · ai_sdk_cohere · ai_sdk_groq · ai_sdk_mistral · ai_sdk_ollama · ai_sdk_mcp · ai_sdk_flutter_ui

See each package's CHANGELOG.md for full details.

Full Changelog: 1.1.0...1.2.0

Errors

• Provider HTTP failures now surface as a typed AiApiCallError carrying the provider's message / type / code / statusCode (and the raw response body), instead of a raw DioException. The AiSdkError hierarchy was relocated into ai_sdk_provider so all provider packages can throw it; ai_sdk_dart's public surface is unchanged via a re-export shim, and existing imports keep working.

Bug Fixes

• onAbort now interrupts streaming. Cancelling the CancellationToken during a streamText run stops the read loop immediately, instead of firing the callback while the underlying stream kept draining to completion.
• stopWhen can extend the tool loop past maxSteps. Custom stop conditions are now the authoritative loop terminator: a higher stepCountIs(...) (or any custom condition) is honored even when maxSteps alone would have stopped earlier, matching Vercel AI SDK v6 semantics. A 1000-step safety cap still backstops runaway loops.

Removed (breaking for the unused video API)

• generateVideo, AiNoVideoGeneratedError, the provider registry's video model category, and the mockVideoModelV1 test helper. No provider ever implemented VideoModelV1 (also removed from ai_sdk_provider), so this API could not be exercised. If you referenced any of these symbols, delete the reference — there was no working code path behind them.

Internal

• Extracted a shared tool-loop run policy (resolveStopConditions, resolveStepBudget, shouldStopAfterStep) and a completeIfPending fan-out helper, de-duplicating generateText and streamText. No public API change.

Coverage

• ai_sdk_dart now has 100% line coverage of lib/.

What's Changed

• Development environment setup: fix dependency conflict + add AGENTS.md by @codenameakshay in #1
• feat: update v1.1.0 by @codenameakshay in #2

New Contributors

• @codenameakshay made their first contribution in #1

Full Changelog: 1.0.0+1...1.1.0

New APIs

• embedMany() — batch embedding that fans out to the provider's doEmbed in configurable chunks, with merged usage stats.
• wrapEmbeddingModel() — middleware pipeline for embedding models, mirroring wrapLanguageModel for language models.
• customProvider() — on-the-fly provider construction from plain model-factory maps without a full createProviderRegistry setup.
• onAbort callback in streamText — fires synchronously when the caller's CancellationToken is cancelled, enabling graceful cleanup.
• timeout parameter — Duration? timeout added to generateText, streamText, generateObject, streamObject, embed, embedMany, generateImage, generateSpeech, transcribe, rerank. Wraps each underlying model call with Future.timeout.
• smoothStream delayInMs option — smoothStream(delayInMs: 20) adds per-chunk delay; experimentalTransform type is now Stream<String> Function(Stream<String>) (async), enabling arbitrary async transforms.
• generateObject / streamObject pass outputSchema — LanguageModelV3CallOptions.outputSchema is now set to schema.jsonSchema so providers that support native structured output (e.g. OpenAI response_format: json_schema) use it automatically.

New Error Types

• AiToolCallRepairError — thrown when a tool-call repair attempt fails.
• AiNoImageGeneratedError — thrown when generateImage receives no image content.
• AiNoVideoGeneratedError — thrown when generateVideo receives no video content.
• AiNoSpeechGeneratedError — thrown when generateSpeech receives no speech content.
• AiNoTranscriptGeneratedError — thrown when transcribe receives no transcript.
• AiRetryError — wraps the list of errors from all exhausted retry attempts.
• AiDownloadError — thrown when an attachment download fails.

Registry Expansion

• createProviderRegistry now supports 6 model categories: language, embedding, image, speech, transcription, and rerank models (previously language + embedding only).
• RegistrableProvider exposes imageModelFactory, speechModelFactory, transcriptionModelFactory, and rerankModelFactory factories.

Testing Utilities

• MockEmbeddingModelV3 — fake embedding model for conformance tests, mirroring MockLanguageModelV2.

Bug Fixes

• pruneMessages — removed duplicate messages that were incorrectly appended in multi-step loops when messages was pre-populated; semantics now match Vercel AI SDK v6.

New Provider Packages

• ai_sdk_azure — Azure OpenAI provider (AzureOpenAIProvider). Language models and embeddings via Azure-hosted deployments; configurable endpoint, apiKey, and apiVersion.
• ai_sdk_cohere — Cohere provider (cohere). Language models, text embeddings, and reranking (Command R, Command R+, embed-english-v3.0, rerank-english-v3.0).
• ai_sdk_groq — Groq provider (groq). Ultra-low latency inference for Llama, Mixtral, Gemma, and other Groq-hosted models.
• ai_sdk_mistral — Mistral AI provider (mistral). Language models and embeddings (Mistral Large, Mistral Small, Codestral, mistral-embed).
• ai_sdk_ollama — Ollama provider (ollama). Local inference for any model pulled via Ollama; no API key required.

Test Count

• 562+ tests covering all public APIs, conformance cases, and new features.

First stable release. Package renamed from ai → ai_sdk_dart to avoid conflicts with existing pub.dev packages.

Core APIs

• generateText / streamText with full v6 result parity — text, finishReason, usage, steps, toolCalls, toolResults, reasoning, sources, files, providerMetadata, request/response envelopes.
• generateObject / streamObject with Output.text/object/array/choice/json API.
• embed / embedMany with cosineSimilarity.
• generateImage, generateSpeech, transcribe, rerank.

Agentic Tools

• Multi-step tool loop with maxSteps, stopConditions, and prepareStep.
• Typed tool definitions (tool<INPUT, OUTPUT>, dynamicTool, Schema<T>).
• Tool choice controls (ToolChoiceAuto, ToolChoiceRequired, ToolChoiceNone, ToolChoiceSpecific).
• Tool approval flow (needsApproval) and ToolLoopAgent for autonomous agentic loops.

Middleware & Registry

• Middleware system: wrapLanguageModel, extractReasoningMiddleware, extractJsonMiddleware, simulateStreamingMiddleware, defaultSettingsMiddleware, addToolInputExamplesMiddleware.
• createProviderRegistry — resolve models by 'provider:modelId' string at runtime.

Streaming

• smoothStream / experimentalTransform stream transform hooks.
• Full fullStream event taxonomy — 22 typed StreamTextEvent subtypes.
• Lifecycle callbacks: onFinish, onStepFinish, onChunk, onError, experimentalOnStart, experimentalOnStepStart.

Other

• Complete AiSdkError sealed class hierarchy.
• Utilities: generateId, createIdGenerator, simulateReadableStream.
• 230-test conformance suite covering all public APIs and provider wire formats.

Full Changelog: https://github.com/codenameakshay/ai_sdk_dart/commits/1.0.0+1

• Improved pubspec descriptions for better pub.dev discoverability.
• Added example/example.md with usage examples and links to runnable apps.