v8.2.2

8.2.2

• Assistant (Beta) feature is now available in the main package. Be aware there might still be bugs due to the beta status of the feature and the SDK itself. Please report any issues you encounter.
• Use "UseBeta": true in your config file or serviceCollection.AddOpenAIService(r => r.UseBeta = true); or new OpenAiOptions { UseBeta = true } in your service registration to enable Assistant features.
• Expect more frequent breaking changes around the assistant API due to its beta nature.
• All Assistant endpoints are implemented except for streaming functionality, which will be added soon.
• The Playground has samples for every endpoint usage, but lacks a complete implementation for the Assistant APIs. Refer to Assistants overview - OpenAI API for more details.
• Special thanks to all contributors for making this version possible!

Other Changes:

• Fixed a bug with multiple tools calling in stream mode.
• Added error handling for streaming.
• Added usage information for streaming (use StreamOptions = new(){IncludeUsage = true,} to get usage information).
• Added timestamp_granularities[] for Create transcription to provide the timestamp of every word.

What's Changed

• WIP Assistant Api by @kayhantolga in #531
• Re-added the beta assistants API by @CongquanHu in #470
• readme update by @kayhantolga in #534
• 8.2.0-beta Assistant by @kayhantolga in #535
• readme update by @kayhantolga in #543
• Feature/assistant api part 2 by @kayhantolga in #546
• Update 8.2.1 to working with ToolCalls and RunSteps by @pappde in #549
• RunRetrieve - add RunStatus.Incomplete by @pappde in #548
• Assistant Vector Support, Message Delete Support, Documentation support. by @kayhantolga in #551
• playgorund cleaned up by @kayhantolga in #559
• gpt-4o Support by @derekforeman in #555
• Streaming fixes. by @SebastianStehle in #558
• Implemented support for Timestamp Granularity on Audio Transcripts by @rjeike in #560
• 8.2.2 by @kayhantolga in #561
• Readme update and version bump by @kayhantolga in #562
• 8.2.2 by @kayhantolga in #563

New Contributors

• @CongquanHu made their first contribution in #470
• @pappde made their first contribution in #549
• @derekforeman made their first contribution in #555
• @SebastianStehle made their first contribution in #558
• @rjeike made their first contribution in #560

Full Changelog: v8.1.1...v8.2.2

8.1.1

8.1.1

• Fixed incorrect mapping for batch API error response.

What's Changed

• 8.1.1 by @kayhantolga in #541

Full Changelog: v8.1.0...v8.1.1

v8.1.0

8.1.0

• Added support for Batch API

What's Changed

• Batch API by @kayhantolga in #537

Full Changelog: v8.0.1...v8.1.0

8.0.1

8.0.1

• Added support for new Models gpt-4-turbo and gpt-4-turbo-2024-04-09 thanks to @ChaseIngersol

What's Changed

• sync by @kayhantolga in #525
• yaml update by @kayhantolga in #526
• Dev by @kayhantolga in #527
• Support for gpt-4-turbo-2024-04-09 + gpt-4-turbo models by @ChaseIngersol in #528
• Version bump 8.0.1 by @kayhantolga in #529
• 8.0.1 by @kayhantolga in #530

Full Changelog: v8.0.0...v8.0.1

8.0.0

8.0.0

• Added support for .NET 8.0 thanks to @BroMarduk
• Utilities library updated to work with only .NET 8.0

What's Changed

• Add DotNet 8.0 Support by @BroMarduk in #443
• Update Readme.md - Update API Docs to DNDocs (domain name changed) by @NeuroXiq in #521
• version bump and small bug fixes by @kayhantolga in #523
• v8.0.0 by @kayhantolga in #524

New Contributors

• @NeuroXiq made their first contribution in #521

Full Changelog: v7.4.7...v8.0.0

v7.4.7

7.4.7

• Fixed a bug that Vision API could not be used with Azure OpenAI, Thanks to @yt3trees
• Fixed a bug that was blocking CreateCompletionAsStream on some platforms. #331
• Fixed a bug that was causing an error with multiple tool calls, now we are handling index parameter #493, thanks to @David-Buyer

What's Changed

• Fixed a bug that Vision API could not be used with Azure OpenAI by @yt3trees in #491
• Fix ChatCompletionCreateResponse handling multiple parallel tool call by @David-Buyer in #494
• 331 createcompletionasstream blocking 1 by @kayhantolga in #515
• Version bump for 7.4.7 by @kayhantolga in #517
• 7.4.7 by @kayhantolga in #516

Full Changelog: v7.4.6...v7.4.7

v7.4.6

7.4.6

• Fixed again🥲 incorrect Model Naming - moderation models and ada embedding 2 model

What's Changed

• fixed incorrect model naming by @kayhantolga in #489
• 7.4.6 by @kayhantolga in #490

Full Changelog: v7.4.5...v7.4.6

v7.4.5

• Fixed function calling streaming bugs thanks to @David-Buyer @dogdie233 @gavi @Maracaipe611
• Breaking Change:
  While streaming (CreateCompletionAsStream), there were some unexpected incoming data chunks like :pings or :events, etc. @gavi discovered this issue. We are now ignoring these chunks. If you were using it, you need to set justDataMode to false.

What's Changed

• Fix function call after 7.4.2 update by @Maracaipe611 in #462
• 442 functioncallinghelpers parameterdescription required = false not work by @kayhantolga in #483
• Reassembly context support for tool call response message by @David-Buyer in #463
• Ignore lines that start with : ping by @gavi in #472
• Fix: #468 by @dogdie233 in #469
• Fixed incorrect new model names by @kayhantolga in #486
• v7.4.5 by @kayhantolga in #487

New Contributors

• @Maracaipe611 made their first contribution in #462
• @David-Buyer made their first contribution in #463
• @gavi made their first contribution in #472
• @dogdie233 made their first contribution in #469

Full Changelog: v7.4.4...v7.4.5

7.4.4

7.4.4

• Added support for new models : TextEmbeddingV3Small, TextEmbeddingV3Large, Gpt_3_5_Turbo_0125, Gpt_4_0125_preview, Gpt_4_turbo_preview, Text_moderation_007, Text_moderation_latest, Text_moderation_stable
• Added optinal dimension and encoding for embedding thanks to @shanepowell

What's Changed

• Add dimensions to EmbeddingCreateRequest by @shanepowell in #476
• v 7.4.4 by @kayhantolga in #479
• Version bump for 7.4.4 by @kayhantolga in #480
• 7.4.4 by @kayhantolga in #481

Full Changelog: v7.0.4...v7.4.4

v7.4.3

7.4.3

• Fixed the response format of AudioCreateSpeechRequest.
• Updated Azure OpenAI version to 2023-12-01-preview, which now supports dall-e 3.
• Added the ability to retrieve header values from the base response, such as ratelimit, etc. Please note that this feature is experimental and may change in the future.
• Semi-Breaking change:
  • The SDK will now attempt to handle 500 errors and other similar errors from the OpenAI server. Previously, an exception was thrown in such cases. Now, the SDK will try to read the response and return it as an error message. This change provides more visibility to developers and helps them understand the cause of the error.

What's Changed

• 452 responseformat wrongfully serialized by @kayhantolga in #453
• handle openai internal errors, return header varlues in baseresponse … by @kayhantolga in #454
• updated azure api version to support latest features. closes #318 by @kayhantolga in #455
• version bump 7.4.3 by @kayhantolga in #457
• 7.4.3 by @kayhantolga in #456

Full Changelog: v7.4.2...v7.4.3