Inception
United StatesModel developer
Palo Alto lab behind Mercury, the first commercial diffusion LLMs.
What they do
Inception Labs, founded by Stanford's Stefano Ermon with colleagues from UCLA and Cornell, builds Mercury — diffusion language models that generate tokens in parallel, reaching 1000+ tokens/sec. Mercury 2 is the current generation; backers include Microsoft M12 and NVIDIA.
Models they develop
1 in this directory.