v1.0.1

What's Changed

• feat: support speed limit for uploading apis by @ikesnowy in #178

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

v1.0.0

BREAKING CHANGES

• GetQwen and GetDeepSeek has been removed
• File upload api for qwen-long has been renamed to OpenAiCompatibleUploadFileAsync
• Microsoft.Extensions.AI dependency has been upgrade to 10.5.0
• No longer reference System.Text.Json for net8.0

What's new

SDK

• feat: support OCR models
• feat: support qwen-MT, qwen-character, qwen-doc-turbo and qwen-deep-research models
• feat: support qwen3-omni-30b-a3b-captioner model
• feat: support enable code_interperator plugin configuration
• feat: support streaming tool calls
• feat: support batch apis
• feat: support freshness, assigned_site_list and intention_options in web search options
• feat: support web search options for multimodal models
• feat: support function call for multimodal models.
• feat: support structured output for multimodal models.

M.S.E.AI

• feat: allow pass TextChatMessage or MultimodalMessage directly to the RawRepresentation
• feat: support tool calls for both non-streaming and streaming api
• feat: allow use ReasoningEffort.None to disable thinking

重大更改

• GetQwen 和 GetDeepSeek 系列 API 已被删除
• 用于 qwen-long 的文件上传接口已被重命名为 OpenAiCompatibleUploadFileAsync
• 包依赖的 Microsoft.Extensions.AI 已升级到 10.5.0
• net8.0 下不再显式依赖 System.Text.Json

更新日志

SDK

• 支持 OCR 系列模型，允许调用内置 Task（高精识别、信息抽取、表格解析、文档解析、公式识别、通用文本识别、多语言识别）
• 增强专项模型支持： qwen-MT, qwen-character, qwen-doc-turbo 和 qwen-deep-research
• 添加 qwen3-omni-30b-a3b-captioner 模型支持
• 添加代码执行器选项支持
• 增加联网搜索参数中关于日期、站点以及自然语言指令的支持
• 多模态模型支持结构化输出
• 增加流式工具调用的支持
• 支持批量推理的创建与管理

M.S.E.AI

• 允许通过 RawRepresentation 直接传递 TextChatMessage 和 MultimodalMessage
• 支持流式工具调用
• 允许通过 ReasoningEffort.None 来关闭思考模式

Full Changelog: v0.9.0...v1.0.0

v1.0.0-pre2

What's Changed

• feat: add qwen audio sample by @ikesnowy in #151
• feat: support sending raw message directly via M.E.AI by @ikesnowy in #156

Full Changelog: v1.0.0-pre1...v1.0.0-pre2

v1.0.0-pre1

What's Changed

• docs: update intl api address by @ikesnowy in #130
• feat: add sample for uploading video file by @ikesnowy in #131
• feat: add ocr sample by @ikesnowy in #135
• chore: remove obsolete apis by @ikesnowy in #136
• feat: add gui sample by @ikesnowy in #137
• feat: support upload files api by @ikesnowy in #139
• feat: support doc url by @ikesnowy in #141

Full Changelog: v0.9.0...v1.0.0-pre1

v0.9.0

What's Changed

• refactor: remove FluentAssertions by @ikesnowy in #111
• Bump Microsoft.Extensions.AI from 9.7.0 to 9.8.0 by @dependabot[bot] in #115
• Bump Microsoft.Extensions.AI.Abstractions from 9.7.0 to 9.8.0 by @dependabot[bot] in #116
• Bump Microsoft.Extensions.AI from 9.8.0 to 9.9.0 by @dependabot[bot] in #120
• Bump Microsoft.Extensions.AI.Abstractions from 9.8.0 to 9.9.0 by @dependabot[bot] in #121
• feat: support temp oss upload by @ikesnowy in #123
• Bump Microsoft.Extensions.AI from 9.9.1 to 9.10.0 by @dependabot[bot] in #125
• feat: update parameters by @ikesnowy in #129
• Bump Microsoft.Extensions.AI from 9.10.0 to 9.10.2 by @dependabot[bot] in #128
• Adds thinking support to multimodal parameters
• Support domain models like roleplay(qwen-character), translation, deep research, data mining
• Support change strategy for web search plugin
• Obsolete QwenCompletion and DeepSeekCompletion overloads

Full Changelog: v0.8.0...v0.9.0

v0.8.0

BREAKING CHANGES

Shortcut methods for LLama2 and BaiChuan LLM are no longer provided, use GetTextCompletionAsync() and specify model name to continue use.

Highlights

TTS support

Usage:

using var tts = await dashScopeClient.CreateSpeechSynthesizerSocketSessionAsync("cosyvoice-v2");
var taskId = await tts.RunTaskAsync(
    new SpeechSynthesizerParameters { Voice = "longxiaochun_v2", Format = "mp3" });
await tts.ContinueTaskAsync(taskId, "博客园");
await tts.ContinueTaskAsync(taskId, "代码改变世界");
await tts.FinishTaskAsync(taskId);
var file = new FileInfo("tts.mp3");
using var stream = file.OpenWrite();
await foreach (var b in tts.GetAudioAsync())
{
    stream.WriteByte(b);
}

More text generation parameters

Support TranslationOptions and SearchOptions in text completion.

Support Logprobs in text completion.

What's Changed

• docs: add sdk issue template by @ikesnowy in #102
• feat: support cosy voice by @ikesnowy in #106
• feat: remove baichuan and llama2 methods by @ikesnowy in #108
• feat: add text generation parameters by @ikesnowy in #109
• feat: update readme doc and samples by @ikesnowy in #110

Full Changelog: v0.7.6...v0.8.0

v0.7.6

What's Changed

• fix: require system.text.json 8.0 for compatibility issue by @ikesnowy in #101

Full Changelog: v0.7.5...v0.7.6

v0.7.5

QWen3 Support

Use TextGenerationParameter.EnableThinking to toggle reasoning on qwen3

What's Changed

• docs: update README by @ikesnowy in #97
• feat: support reasoning switch for qwen3 models by @ikesnowy in #100

Full Changelog: v0.7.4...v0.7.5

v0.7.4

BREAKING CHANGES

• The minimal framework requirement has dropped to net6.0, there might be framework-related issues, but should be ok for most of times, just a reminder if wired bug happens.
• Cnblogs.DashScope.Sdk no longer depends on JsonSchema.Net.Generation, you may manually install it or use GetJsonSchemaAsNode() if you are in net9.0(docs).

What's Changed

• chore: upgrade json schema by @ikesnowy in #92
• build(deps): Update to M.E.AI 9.5.0-preview by @jozkee in #93
• chore: upgrade msai 9.5.0 by @ikesnowy in #94
• refactor: make library compatible with net6 by @ikesnowy in #96

Full Changelog: v0.7.3...v0.7.4

v0.7.3

What's Changed

• Update to M.E.AI 9.4.3-preview.1.25230.7 by @jozkee in #91

Full Changelog: v0.7.2...v0.7.3