LFM2.5-24B and Qwen3.6-35B on llama.cpp on a 6GB laptop GPU, driven by pi, Codex and Claude Code. The agent's own system prompt decided the result.
Same Local Model, 3 Coding Agents: pi 12/12, Codex 12/12, Claude Code 7/12 on a 6GB GPU
Related
I Built a Latent Diffusion Model That Fits in 5.5MB and Runs Without a GPU
TinyDiffusion 128 is a 1.45M-parameter Diffusion Transformer that runs on a plain CPU. Here are the ablations, benchmarks, and the caveats.
Which model, settings and prompt? Testing an LLM step one change at a time
Most write-ups about AI automation show what was built. Fewer show how the builder knows it can be...
Why your local LLM tok/s numbers don't match mine
A few weeks ago I needed to pick a local inference server for a project. I'm on a Mac, so the...