LegacyRemaster

• Waiting for Qwen 3.7 open weight... The new King has arrived...

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 282 comments

• Testing MiMo-V2.5-IQ3_S with 1'048'576 context

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 17 comments

• New models when? Forecasting release date.

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 84 comments

• Trying Mimo 2.5 pro with RTX 6000 96gb + W7800 48gb + 128gb DDR4

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 7 comments

• Cuda + ROCm simultaneously with -DGGML_BACKEND_DL=ON !

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 27 comments

• XiaomiMiMo MiMo-V2.5 (not pro) - Architecture: Sparse MoE (Mixture of Experts), 310B total / 15B activated parameters

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 16 comments

• Is it possible to edit LLAMA.CPP with Cline+Vscode+Minimax 2.7 Q4_K_S and get a working build?

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 13 comments

• Where we are. In a year, everything has changed. Kimi - Minimax - Qwen - Gemma - GLM

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 30 comments

• Minimax 2.7: good news!

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 41 comments

• FT - China’s Alibaba shifts towards revenue over open-source AI

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 132 comments

• Qwen 3.5 397B vs Qwen 3.6-Plus

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 74 comments

• We have a new weight class...

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 123 comments

• Minimax 2.7: Today marks 14 days since the post on X and 12 since huggingface on openweight

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 81 comments

• I understand the disappointment if minimax 2.7 does not become open weights but we have had a lot..

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 28 comments

• Trellis 2 run locally: not easy but possible

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 22 comments

• Minimax 2.7: If the model updates itself, who is responsible for the behavior of the version running in production?

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 7 comments

• Claude Code with LM studio: 0.4.1

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 13 comments

• Testing Fine-tuning Studio

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 1 comments

• Is memory speed everything? A quick comparison between the RTX 6000 96GB and the AMD W7800 48GB x2.

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 18 comments

• Model: support GLM-OCR merged! LLama.cpp

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 7 comments

• The impossible marriage: GLM 5 unsloth TQ1 test on 1xRTX 6000 + 2xW7800 48gb.

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 3 comments

• I tested qwen next 80b instruct with Claude Code on llama.cpp. I created the Tetris game and it worked 100% locally.

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 46 comments

• Qwen3.5-397B-A17B : a significant step forward in many benchmarks but still too many hallucinations

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 16 comments

• Qwen3.5-397B-A17B will be open source!

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 27 comments

• From Minimax: Forge: Scalable Agent RL Framework and Algorithm

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 1 comments

• Ring-1T-2.5 Trillion Parameter Open Source Thinking Model

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 0 comments

• While we wait for Deepseek 4, Unsloth is quietly releasing gguf for 3.2...

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 12 comments

• Cerebras MiniMax-M2.1-REAP-139B-A10B - Mradermacher Q4_K_S tested

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 23 comments

• Scrolling through the trending list on huggingface I found LightOnOCR-2-1B ....

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 4 comments

• Model: cerebras/GLM-4.7-REAP-268B-A32B incoming!

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 50 comments

• OpenAI has signed a $10 billion contract with Cerebras

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 37 comments

• Tested Glm-4.7-REAP-40p IQ3_S . Single RTX 6000. Works

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 15 comments

• XiaomiMiMo.MiMo-V2-Flash: is there a reason why i see so few ggufs?

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 36 comments

• KT-Kernel achieves up to >4.5x prefill and 30% faster decode compared to llama.cpp on the same hardware , why?

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 8 comments

• Minimax M2.1 is out!

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 47 comments

• I was waiting for Minimax and MiMo-V2-Flash arrived!!!

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 15 comments

• ik_llama GLM 4.7 : 8~9 tokens/sec (ubergarm) instead of 4.5~5 tokens/sec (llama.cpp)

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 6 comments

• Qwen3-Next-80B-A3B-Thinking-GGUF has just been released on HuggingFace

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 52 comments

• We basically have GLM 4.6 Air, without vision

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 14 comments

• I just wanted to do a first benchmark of GLM 4.6 on my PC and I was surprised...

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 35 comments

• MiniMax-M2 Asteroid game - Unsloth

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 9 comments

• Minimax M2 thinks it's GPT...

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 4 comments

• I'm testing the progress on GitHub. Qwen Next gguf. Fingers crossed.

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 15 comments

• 8 GPU Arc Pro B60 setup. 192 gb Vram

  Posted by LegacyRemaster@reddit | LocalLLaMA | View on Reddit | 14 comments