Skip to content

Releases: unkn0wncode/openai

v1.4.0

Choose a tag to compare

@unkn0wncode unkn0wncode released this 19 May 14:17

What's Changed

Breaking Changes

  • Reworked Responses streaming iteration API (3efbd81)
    • Replaced stream.Chan() with stream.Seq() for (event, err) range loops
    • stream.All() now returns ([]any, error)
    • Stream error events are available through stream.Err() / StreamError
  • Renamed model constants to follow model IDs more closely (97cfc68)
  • Made tool registry locking internal instead of exposing the embedded mutex (97cfc68)

Features

  • Added web search tool options (da1bb7f)
    • Added domain filters, return token budget, and external web access controls

Improvements

  • Improved Responses SSE and WebSocket stream cleanup and error handling (3efbd81)
  • Increased background polling timeouts for slower background responses (80177e0, 8a4c3c8)
  • Fixed a data race when registering tools (97cfc68)

Models

  • Added gpt-5.3-chat-latest, gpt-5.4, gpt-5.4-mini, gpt-5.4-nano, gpt-5.4-pro, gpt-5.5, gpt-5.5-pro, chat-latest, gpt-image-2, gpt-realtime-2
  • Added duration-billed realtime model pricing data (633f7b0)
  • Updated default model to GPT-5.5 and default image model to gpt-image-2
  • Removed dropped series 4 and GPT-4o audio/realtime preview models (9542c4c, 71ca57b)

Full Changelog: v1.3.1...v1.4.0

v1.3.1

Choose a tag to compare

@unkn0wncode unkn0wncode released this 02 Mar 12:22

What's Changed

  • Added support for server-side context compaction via responses.Request.ContextManagement (5bc451f)
    • New output.Compaction type for compaction items returned by the API

Full Changelog: v1.3.0...v1.3.1

v1.3.0

Choose a tag to compare

@unkn0wncode unkn0wncode released this 25 Feb 19:06

What's Changed

Features

  • Added WebSocket support for Responses API streaming (7e65a5c)
    • client.Responses.WebSocket opens a persistent connection; WSConn.Send returns the same StreamIterator as SSE-based Stream
    • Includes Warmup for pre-heating request state without generating output
  • Added support for tool CallLimit option in Responses API (95a8362)

Fixes

  • Fixed unmarshaling error due to overlap between fields of computer use and shell tools (96f39d1)

Models

  • Added gpt-5.3-codex, gpt-realtime-1.5, gpt-audio-1.5
  • Removed gpt-4o-latest, codex-mini-latest after they were dropped from the API

Full Changelog: v1.2.1...v1.3.0

v1.2.1

Choose a tag to compare

@unkn0wncode unkn0wncode released this 17 Jan 14:37

What's Changed

New

  • Added gpt-5.2-codex model (f1cfb3a)
    • New constant GPT52Codex for the model
    • Added pricing data matching official pricing page

Full Changelog: v1.2.0...v1.2.1

v1.2.0

Choose a tag to compare

@unkn0wncode unkn0wncode released this 08 Jan 19:37

What's Changed

Features

  • Added support for GPT-5.2 model series (8e5c988)
    • New models: gpt-5.2, gpt-5.2-pro, gpt-5.2-chat-latest
    • Updated default model to GPT-5.2
  • Added support for persistent conversations (8ae6b0f)
    • New conversation management API for multi-turn interactions
    • Added example for conversation usage

Improvements

  • Updated models and pricing to match the API (551604c)
    • Added new image models: chatgpt-image-latest, gpt-image-1.5
    • Added new TTS models: gpt-4o-mini-tts-2025-03-20, gpt-4o-mini-tts-2025-12-15
    • Added new text models: gpt-4o-mini-transcribe-2025-03-20, gpt-4o-mini-transcribe-2025-12-15, gpt-realtime-mini-2025-12-15, gpt-audio-mini-2025-12-15
    • Updated pricing for new models based on official pricing page
  • Updated list of models (cf620b4)

Chores

  • Deleted unused test file (203c713)
  • Excluded .github directory from gitignore (d0bfe79)

Full Changelog: v1.1.0...v1.2.0

v1.1.0

Choose a tag to compare

@unkn0wncode unkn0wncode released this 16 Nov 10:37

What's Changed

Features

  • Added support for GPT 5.1 models
  • Added support for new tools (8aeaef1)
    • New apply_patch tool with examples and helpers
    • New shell tool with examples and helpers
    • Added cache retention parameter prompt_cache_retention
    • Added new effort level none
  • GitHub Actions workflow to check model list regularly (fd53f5f)
  • Added a test to track new/deleted models (9f3cf2f)

Improvements

  • Optimized tests (f325f31)
    • Tests now run in parallel
    • Use lower reasoning for faster execution
    • Disabled deprecated Assistants test
  • Updated models/reasoning in slow examples (2865c55)

Fixes

  • Updated lists of models to match the API (32661d2)
    • Added separate lists for Video and TTS models
    • Removed mentions of deleted models from the codebase
    • Updated test for missing models with better output
  • Fixed incorrect name handling when registering built-in tools (8aeaef1)
  • Unnamed protective mutex in config (a828277)

Chores

  • Removed dead safety code (a294445)
  • Updated checkout action version (b5cc6fd)

Full Changelog: v1.0.2...v1.1.0

v1.0.2

Choose a tag to compare

@unkn0wncode unkn0wncode released this 15 Nov 12:13

What's Changed

  • Updated response struct to parse instructions of varied types

Full Changelog: v1.0.1...v1.0.2

Add custom tools

Choose a tag to compare

@unkn0wncode unkn0wncode released this 24 Aug 16:00
5df4316

What's Changed

  • feat: add custom tools support by @unkn0wncode in #2
  • fix: verbosity param moved to text.verbosity

Full Changelog: v1.0.0...v1.0.1

v1.0.0

Choose a tag to compare

@unkn0wncode unkn0wncode released this 02 Aug 18:57

Initial stable release