Inception released Mercury 3, a diffusion text model. First one I've seen where "smarter" and "faster" both actually hold up. It beats GPT-6 Luna and Claude Haiku 5.5 on benchmarks. 1068 tok/s, which is absurd. 1M context, $0.10 in and $0.35 out per million tokens.
Inception released Mercury 3, a diffusion text model. Inception released Mercury 3, a diffusion text model.