Redditr/LocalLLaMA07:40RTX 5070 Ti + 9800X3D running Qwen3.6-35B-A3B at 79 t/s with 128K context, the --n-cpu-moe flag is the most important part.中文翻译暂不可用
Redditr/LocalLLaMA06:37qwen3.6 performance jump is real, just make sure you have it properly configured中文翻译暂不可用
Redditr/ClaudeAI05:17Got a 3D printer for the team, accidentally started a Claude Code crab invasion中文翻译暂不可用
XKimi.ai03:22We don't know what Kimi + Hermes agents will look like in the wild yet. That's why we built the hackathon. Presented by @Kimi_Moonshot & @NousResearch...中文翻译暂不可用
Redditr/LocalLLaMA02:55Accidentally discovered you can teach frozen MoE models new knowledge by just steering their expert routing — no training needed中文翻译暂不可用
XElon Musk02:39RT Ethan He: grok speech-to-text and tts apis are out. multilingual multi-speaker is huge for me personally. https://x.ai/news/grok-stt-and-tts-apis中文翻译暂不可用