The exhaustive, unofficial Dart bridge to OpenAI

What it does

dart_openai is a community-built wrapper that maps OpenAI’s REST endpoints into Dart-typed methods. It covers the bulk of the current API catalog—chat responses, image generation, audio transcription, embeddings, evals, vector stores, and more—so Dart and Flutter developers can call GPT-4, DALL-E, and Whisper without writing raw HTTP boilerplate.

The interesting bit

The package treats OpenAI’s documentation as a spec sheet and systematically ticks it off: the README includes a coverage table that tracks completeness down to individual sub-endpoints like conversation items and eval runs. That bureaucratic thoroughness is actually the selling point—it is trying to be the reference implementation OpenAI never provided.

Key highlights

• Covers most major APIs: Responses, Conversations, Audio, Images, Embeddings, Files, Models, Moderation, Vector Stores, and Containers are marked complete.
• Explicitly skips beta features like Assistants and ChatKit, so you won’t find half-baked implementations of unstable endpoints.
• Includes typed support for niche endpoints like Evals, Graders, and Fine-tuning (the latter is currently at 70% coverage).
• Optimized for Flutter with file-based helpers for image edits and audio speech output.

Caveats

• Several newer or streaming features are still on the roadmap: Videos, Real-time, Batch, Uploads, and Image Streaming are listed as planned but not yet implemented.
• Chat completion streaming is notably excluded from the “complete” checklist, despite the feature list claiming general streaming support.
• The README has visible rough edges—typos like “Images Streaaming” and inconsistent date formatting—that suggest maintenance is functional but not polished.

Verdict

Dart and Flutter developers who want a typed, batteries-included client for standard OpenAI tasks should look here. If you need bleeding-edge features like real-time APIs or Assistant threads, you’ll still be waiting or writing raw HTTP.