Geeky Uncle

💡每天5分钟，轻松背单词 📚 四六级 | 考研 | 雅思 | 留学 🎯 实用英语 | 必备干货 | 刚需场景 ✨ 高效记忆，单词稳稳拿捏！

[Geeky Uncle @GeekyUncle]

@ivanfioravanti What do you think ds4-agent vs pi + oMLX?

[Geeky Uncle @GeekyUncle]

@GoSailGlobal @grok 原始视频在哪里

[Geeky Uncle @GeekyUncle]

@lmstudio nice！expecting 122b

[Geeky Uncle @GeekyUncle]

@grok @aliez_ren @grok 为什么我在grok里面问，回答我说M4 Max支持MXFP4？

[Geeky Uncle @GeekyUncle]

@aliez_ren @grok MLX不是支持MXFP4吗？

[Geeky Uncle @GeekyUncle]

@Prince_Canuma I failed at downloading the models. I noticed that it always downloads the full size model weight, plus fp8 and fp4 versions. Is it possible to have a parameter so that I can only download fp4 version to have a try?

[Geeky Uncle @GeekyUncle]

codex "hi" --oss --local-provider lmstudio
"error": {
"message": "Invalid discriminator value. Expected 'function' | 'mcp'",
"type": "invalid_request_error",
"param": "tools.6.type",
"code": "invalid_union_discriminator"
}
Codex v0.87.0 && LMStudio v0.3.39
@lmstudio

[Geeky Uncle @GeekyUncle]

@Prince_Canuma Really amazing! I will try on my M3 Max!

[Geeky Uncle @GeekyUncle]

@CulStory compare to GPU version，what's the difference?

[Geeky Uncle @GeekyUncle]

@harshildesigner @figma Figma

[Geeky Uncle @GeekyUncle]

有没有人和我一样喜欢这种趣味学英语的方式？每天打卡一个世界著名地标？一边学英语，一边看世界？

[Geeky Uncle @GeekyUncle]

@Prince_Canuma May LMStudio integrates MLX-audio for voice chat? @lmstudio @lmstudiodevs

[Geeky Uncle @GeekyUncle]

@Prince_Canuma Wow! I see Qwen Omni!

[Geeky Uncle @GeekyUncle]

anyone use LMStudio response API with JSON schema? @lmstudiodevs @lmstudio

[Geeky Uncle @GeekyUncle]

@Prince_Canuma @minimax_ai great！the output is really verbose

[Geeky Uncle @GeekyUncle]

@Prince_Canuma Congrats!

[Geeky Uncle @GeekyUncle]

@ivanfioravanti @filipstrand Qwen Image Edit include?

[Geeky Uncle @GeekyUncle]

@Prince_Canuma That’s awesome