Many of you have written to me: "but which model can I download with Ollama and run on my computer?"
Models like the GLM 5.3, Qwen 3.8 or Kimi K3 have a quality comparable to that of the frontier models. Except they don't run on the computer. GLM 5.3 has 753 billion
A GPU that almost everyone has, did this without a single line of handwritten code.
An RTX 3060, the most common GPU on Steam, built an entire game. → Bonsai2, a Qwen 3.8 27B model, running on 12GB of VRAM. → 5 hours, 328 thousand tokens, 2,368 lines. The GPU
Qwen 3.8 Flash decisively crushes the M5 Ultra with this new recipe by vr8vr8 .
I think I found the fastest dual DGX Spark recipe for Qwen3.8-Flash-Next. 138 tok/s structured, 133 code, 383 structured aggregate at four streams, and now it talks to Hermes. It feels like a
Voxel demos on X are all paint and no thought.
A model renders a pretty pagoda and we call it smart. So I built a challenge instead. Same prompt, three local models so far, one self-contained HTML file each. The challenge Build a bridge over a river. Exactly 40 blocks, 4
You have reached the end of the archive
All of qwen38