NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
pub.dev · #2640 most downloaded on pub.dev
Unofficial Dart & Flutter SDK for the OpenAI API. Typed clients for Responses, Chat, Realtime, Videos, Batch, Fine-tuning, Vector Stores and more.
Last release 1 months ago
25 Aug 2026
Ships unpredictably
gaps range from 9 days to 1.7 years
Nearly every release is documented
notes for 51 of 55 stable releases
Nothing withdrawn
no release was ever pulled
4 years old
55 releases · first in 2023
One column per quarter.
The package now compiles and runs on Flutter Web. Every unconditional dart:io import is gone from lib/ ; file uploads are handled through a platform-n
dart:io import is gone from lib/; file uploads are handled through aOpenAIFile(bytes:, fileName:) type, and disk helpers areOpenAIRetryPolicy, on by default): GET/DELETE retryRetry-After wins.maxRetries: 1.StreamTimedOutException instead of hanging forever.x-ratelimit-* headers are parsed after everyOpenAIResponseMeta.lastRateLimit.UnimplementedError.RequestFailedException.final azure = OpenAIClient(
apiKey: '<key>',
azure: const OpenAIAzure(
resource: 'my-resource',
apiVersion: '2024-10-21',
deployments: {'gpt-4o': 'gpt4o-prod'},
),
);Requests are rewritten to /openai/deployments/{deployment}/...?api-version=
transparently.
OpenAIFile instead of dart:io File.await loadOpenAIFile('path') on native platforms or constructOpenAIFile(bytes: ..., fileName: ...) anywhere.audio.createSpeech returns Uint8List bytes (optionally writing to diskoutputDirectory/outputFileName are provided as strings).Support for containers & container files
Support for all Vector Store APIs
Fixed Image Edit's image mime type issue.
[Audio] createTranscription Added timestamp_granularities[] request parameter and updated response object by @mihirshahxenlabs in #157
timestamp_granularities[] request parameter and updated response object by @mihirshahxenlabs in #157/chat body compatibility by @lollipopkit in #178Full Changelog: v5.0.0...v6.0.0
Complete rewrite, retest and implementation for the package to match latest API changes from OpenAI including all new features and endpoints.
Bug fixes.
Massive issues fixes related to incorrecr use of types, model fields..., audio transcription file streaming error handling, and more.
Massive issues fixes related to incorrecr use of types, model fields..., audio transcription file streaming error handling, and more.
Support for newer fields & options for the chat completions API, like name for messages.
Support for audio transcription granularity.
General improvements, and more tests.
Notifications about breaking changes in the chat completions methods in favor of OpenAI Vision feature support, JSON mode..
RequestFunctionMessage's toMap() method.Removed the exposed field for configuring the package to use fetch_client instead of http_client manually withe is isWeb field, in favor of using dart
isWeb field, in favor of using dart.library.js and dart.library.io conditional imports to automatically detect the platform and use the appropriate client for it.Migrated for the deprecated \_functions\_and function_call th the use of new fields such tools.. in the chat API, confirmed that it works on both asyn…
functions_and function_call th the use of new fields such tools.. in the chat API, confirmed that it works on both asynchronous and stream responses.Disabled a print method that was used for debugging purposes.
Install it in your Dart/Flutter app from:
Fixed the non applied n field in image variation method.
n field in image variation method.model field for other image APIs methods.Install it in your Dart/Flutter app from:
Followed up with more changes in the OpenAI API.
Migrated internal packages to latest versions such as http.
Added support for functions feature of the chat API.
- Formated unformatted files.
Exposing of custom thttp client for methods- Fixed issue related to streams that uses a custom http client.
Added custom HTTP client options for all the package APIs that requires one.
exposed an external method for inclusing custom headers to the requests.
- Fixed minor issue.
More refatocration for the packages services.
Fixed the non catched error for stream functionality of chat completions and completions.
Added more examples for the package use in /example folder.
Added more helper memebers for some models.
- More issues fixes.
Nothing published for this version
Exposed more enums to be used with the library.
base url not changeable after first request fixed.
Fixed web issue for chat streams
Added the Audio APIs including creating transcription and translation.
Nothing published for this version
Nothing published for this version
Added documentation of chat completion stream.
Added And Fixed chat completions stream.
Added new ChatGPT API to the library.
Ensured immutability of results from requests.
Fix doubled stream snapshots than expected.
Changed the use of dotenv to envied package.
Fixed echo property type misleading.
Added implementation of fine-tune events Stream.
Fixed the Stream based mechanism for completion stream.
Exposed image size and url enums to be used externally
Nothing published for this version
- formatted some dart files
Added more detailed documentation that reflects OpenAI's.
Added more detailed documentation that reflects OpenAI's.
Set internal packages version to any
Downgraded internal meta package version to 1.8.0
Added example to showcase how to use.
Informed about using the dotenv package instead of flutter_dotenv since not all dart applications are flutter's.
- added tests for fine-tunes - fixed type system issues -
Added error handling explanation, example in the README.md
Added clarifications in READMe.md
- Fixed issues
Added documentation for responses fields for better understanding, familiarization for developer.
- All APIs are included.
Your coding agent can read these notes before it upgrades. Set up the MCP server →