Someone got DeepSeek-V4.1-Flash running at 38.2 tps on an AMD Strix Halo. 🤯
This is a AI post classified by AGK Intelligence as Open & local models (a model release), kept by the AI Radar because it carries real work, not commentary.
Someone got DeepSeek-V4.1-Flash running at 38.2 tps on an AMD Strix Halo. 🤯 No multi-GPU server, just a single Ryzen AI Max+ 395 with 128GB unified memory. why should I care? ... because this model is enormous Setup 🧠 552B backbone parameters 💾 ~196B additional Engram parameters 🔥 16B active during decoding 🖥️ AMD Ryzen AI Max+ 395 🎮 Radeon 8060S integrated GPU 💾 128GB unified memory 🐧 Linux + ROCm 7.x The secret? YES! You are correct, new inference engine!! A new open-source inference engine called libosmos. Developer-reported benchmarks using Unsloth's UD-IQ1_S quantization: ⚡ 38.2 tps
Posted by David Hendrickson (12.8k followers) 14 h ago · 203 likes · 17.9k views · view the original post on X. Kept by the AI Radar as Open & local models.
More AI work like this
- Qwen 4 (unnamed quant) — @loktar00
- 北冥有鱼，其名为鲲。 — @NFT_Chen
- If you want to get started with Local AI but dunno where to start, THIS IS FOR YOU 🚨 — @javierbits
- 💥重磅！Qwen3.8-Flash-Next 量化版炸裂升级：Token消耗直接腰斩50%，准确率暴涨15%！本地32GB丝滑跑通125B级MoE！ — @NFT_Chen
- This could be so cool. — @niklaslenz_ai
- Qwen is teasing a new model — @RoundtableSpace
- 时代变了，以前是token每秒，有人提出了秒每token。。。。 — @mylifcc
- Qwen3.8-Flash-Next GGUFs now run faster with less thinking and less token use. — @UnslothAI
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 65.6k posts from 5k X accounts over the last 14 days, 10.4k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 16:05 UTC. Full method.