Pinned
joseph
@gitcommit90
nano bio.md
director of @publicbytesorg
idk why it's so funny to me they included the ™ symbol hereQwen3.8 27B brings a new state-of-the-art dense model for local AI development. ⚡ Run it on AMD Ryzen™ AI Max+ processors or single Radeon™ AI PRO R9700 card ⚡ Experience it with @lmstudio ⚡ Turn it into an app with @lemonade_server Start building with AMD Day 0 support for- Unless all of these leaders have unlocked something novel, 10/10 don’t recommend. Can I drive with 10 miles left in the tank? Yes. Will I pass a hundred gas stations? Also yes. Fitting the tokens ≠ using them effectively. “Lost in the middle” has been replicated for years.In my conversations with leaders at OpenAI, Anthropic, and Google, I'm just going to say that they don't seem to worry about context length/context windows. Like, at all. They all have crazy long-running threads. One told me one of his threads has billions of tokens.
- Run @Alibaba_Qwen 3.8 27B on a single @NVIDIAAI DGX Spark 🔥 at 256k context + Vision @ ~22tok/s @UnslothAI NVFP4 · vLLM 0.26.0 · MTP k=3 vision ON · one GB10 21.4 tok/s avg. (20 completed runs, 5x for each effort mode: none, low, medium, xhigh) Concurrent requests: 20.6
- The long awaited @Alibaba_Qwen 3.8 27b is interesting. Not judging a model by its benchmarks, getting it set up on the @NVIDIAAI DGX Spark now, but if these benchmarks hold, it looks like for most GB10 owners, Deepseek V4 flash 0731 is still going to stand. Stay tuned for the 3.8
