20 days ago we built a system that writes GPU code.
Today its kernels outperform the state of the art. We just open-sourced 9 megakernels, which outperform NVIDIA’s stack on DeepSeek and run self-driving, robotics, protein folding and world models 2–3× faster than the current
Catalog note, Built in Venice: the same snapshot picked up DeepSeek V4.1 Flash 1017 Abliterated - E2EE, private, 1M context.
Uncensored and private on the same route.
Two private models, now included with Pro.
DeepSeek V4.1 Flash and GLM 5.3 Flash run in Private mode, so no prompt or response is stored. Try it on Venice.
Baidu Open-Sources ERNIE 5.5 Flash Base, a 280B Model Using 14B per Token
Baidu_Inc released the base weights of ERNIE-5.5-Flash-Base on huggingface. Baidu says it's the pre-training checkpoint, before any
You have reached the end of the archive
All of deepseek