Report
Inception launches Mercury 3, a diffusion LLM it claims generates about 1,000 tokens per second
Techmeme, citing The Deep View, says Inception calls Mercury 3 comparable to Luna and Haiku but 4x to 8x faster.
TLDR
Techmeme, citing The Deep View’s Jason Hiner, reports that Inception launched Mercury 3, a diffusion-based LLM. Inception claims it is comparable to Luna and Haiku models but 4x to 8x faster, generating about 1,000 tokens per second.
Combined views
1.8K
1 Source, first seen ago