Hassan trains Tev1 0.8B, a tiny Jev-like classifier

Hassan says he has trained Tev1 0.8B, a tiny classifier he describes as Jev-like. On his Mac, it runs fully locally through Ollama and classifies tasks.

He puts end-to-end latency for that local run at about 50 ms. The weights and benchmarks are not available yet, and Hassan says he will release them very soon.