LocalLLaMA
StepFun 3.7 Flash MTP Bench Strix Halo
This is the StepFun Step-3.7-Flash UD-IQ4_XS main model with the official StepFun MTP Q8_0 draft model, served through a patched llama.cpp Vulkan/RADV build. Host System: AMD Ryzen AI Max+ 395 / Radeon 8060S ( gfx1151 ) Memory: 128 GB unified LPDDR5X BIOS UMA / VRAM: 4 GB UMA dedicated VRAM GTT ceiling: 112 GiB IOMMU: enabled ( amd_iommu=on ) OS: Ubuntu 25.04 (Plucky) Kernel: 6.18.1-061801-generic Mesa / RADV: Mesa 25.2.8 / RADV ROCm: 7.1.1 baseline; some later ro...
Read the full discussion on Reddit
This article was aggregated from Reddit. Click to join the conversation.
View on Reddit
Related Trends in LocalLLaMA
Building an open-source control plane for self-hosted vLLM, what would you want in it?
5090: Windows or Linux for Qwen3.8.27b
RTX Pro 4500 vs. A100 (x4) - purchase being made today
Best llm at understaning middle eastern languages? Arabic/Persian etc
Qwen3.8-27B abliterated FP8: refusal 64β99% β 0β6%, and MMLU/GSM8K move less than 1.3 points
llama.cpp Windows Manager