Inception Labs' Mercury 2 diffusion language model outperforms Google's DiffusionGemma, generating 1,009 tokens per second while retaining reasoning

Both models trade word-by-word generation for parallel denoising. Only one of them does it without losing intelligence in the trade.

Inception Labs' Mercury 2 diffusion language model outperforms Google's DiffusionGemma, generating 1,009 tokens per second while retaining reasoning

Inception Labs' Mercury 2 AI outperforms Google's DiffusionGemma. Google having the best AI model by June 2026 at 4.6% YES.