Qwen 3.8 Flash Next ROCmi4 built this with OMP on AMD Strix Halo and I am so happy to…
This is a dev post classified by Jev as AI dev tools (a tool drop), kept by the Dev Radar because it carries real work, not commentary.
Qwen 3.8 Flash Next ROCmi4 built this with OMP on AMD Strix Halo and I am so happy to make a fully working quant of this. This works in Hermes + OMP and other coding harnesses. It built this retro cube game PLUS enabled it over my tailscale for me. It's a very capable model. Realistic Prefill: 700-768 tok/s Realistic Decode: 53 tok/s To keep it coherent, without messing up and flooding loops I had to change the original recipe on this one. Initially I was getting 79 tok/s decode with 970+ prefill and that may have worked fine for llama serve but it had some loops and those loops were wors
Posted by Carlo (7.3k followers) 44 min ago · 8 likes · 311 views · view the original post on X. Kept by the Dev Radar as AI dev tools.
More dev work like this
- One of my favorite features in ChatGPT / Codex is also one I rarely see anyone talk… — @Voxyz_ai
- 💥划重点！Claude Opus 5.5 今天上了，这次真给订阅党面子 ！ — @NFT_Chen
- Grok 4.7 just landed on @mercor's APEX leaderboards: — @AlphaSignalAI
- Claude is back. — @Yuchenj_UW
- the "six vendors to run one agent" problem is solved — @Muskanjain0401
- Anthropic is offering a free reset to "Explore Opus 5.5" — @neerajjj6785
- Grok Build just got another major upgrade, expanding what Grok can do far beyond coding — @XFreeze
- Claude Opus 5.5 from @AnthropicAI is live on OpenRouter! — @OpenRouter
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 21.2k posts from 5k X accounts over the last 21 days, 2.4k tools, 12 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-22 17:03 UTC. Full method.