Mr_Moonsilver

• What memory system are you using for your agents?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 61 comments

• I pray there is a Qwen 3.6 122b version (4x3090 owner)

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

• Wen 3.7?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 5 comments

• 4 x 5070 Ti dual slot in one build?

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

• Benchmark Qwen 3.6 27B MTP on 2x3090 NVLINK

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 40 comments

• Still waiting for Grok 3 to go opensource

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 56 comments

• What happened to Deepseek?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 138 comments

• Does anyone here rember EleutherAI with GPT-Neox-20b? Or BigScience Bloom 176B?

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

• GLM releases OCR model

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

• Why has no one been talking about Open Hands so far?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 135 comments

• R9700 users - Which quants are you using for concurrency?

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

• Completed Local LLM Rig

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 156 comments

• Qwen leadership leaving had me worried for opensource - is Nvidia saving the day?

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

• Is anyone using Qwen Next Coder for clawdbot locally?

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

• Qwen 80b next in OpenwebUI misses thinking tags

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

• Has anyone tested the RX 9060 XT for local inference yet?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 21 comments

• Deepseek OCR 2

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

• 1600W enough for 2xRTX 6000 Pro BW?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 29 comments

• AWQ-quantizing Qwen-3-VL-Embedding/Reranker models?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 2 comments

• Has anyone tried the new 'auto' feature for vLLM?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 2 comments

• Running FP8 with vLLM on RDNA4?

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

• Are you using local LLMs to power Cline, RooCode, Cursor etc.? What is your experience?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 31 comments

• Is it just me or has CES really not delivered anything exciting for local LLM setups?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 21 comments

• Meta drops new ASR models (up to 7B)

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

• Memory Bandwith of GPUs, the only defining factor for speed?

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

• Ultra Low Latency PC Project

  Posted by Mr_Moonsilver@reddit | buildapc | View on Reddit | 3 comments

• Will we see: Phi-5, Granite 4, Gemma 4, Deepseek R2, Llama 5, Mistral Small 4, Flux 2, Whisper 4?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 88 comments

• Would it make sense to train a model on Roo Code/Cline?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 5 comments

• What's your hope we still get to see GLM 4.6 Air?

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

• GPT-OSS-120B Performance on 4 x 3090

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

• K2-Think 32B - Reasoning model from UAE

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 49 comments

• The issue with SWE bench

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

• Do you think Qwen3 VL will get a release for other models too?

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

• Strange Sounds from Speakers when GPU-Rig is computing

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

• Insights on performance degradation for Qwen3 30B3A?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 5 comments

• Prompting forces you to have a clear vision (especially w GPT5)

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 5 comments

• Has someone used OWebUi with Docling to talk to pdfs with visualizations?

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

• "AGI" is equivalent to "BTC is going to take over the financial world"

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 331 comments

• OSS release = pressure for xAI to opensource Grok 3?

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

• How is ROCm support these days - What do you AMD users say?

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 78 comments

• InternVL3: Advanced MLLM series just got a major update – InternVL3-14B seems to match the older InternVL2.5-78B in performance

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 24 comments

• OpenAI performs KYC to use the latest o3-pro via API

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 70 comments

• Hardware Repair Shops

  Posted by Mr_Moonsilver@reddit | buildapc | View on Reddit | 0 comments

• Well...

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

• Apple be like...

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

• Speculative Decoding + ktransformers

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

• Google opensources DeepSearch stack

  Posted by Mr_Moonsilver@reddit | LocalLLaMA | View on Reddit | 78 comments

• Custom 12VHPWR to 2x8-Pin: Pins Missing?

  Posted by Mr_Moonsilver@reddit | buildapc | View on Reddit | 33 comments

• With Veo3 producing hyper realistic content - Are we in for a global verification mechanism?

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

• AsRock Wrx80 Pci power delivery

  Posted by Mr_Moonsilver@reddit | buildapc | View on Reddit | 2 comments