BREAKING
Inception Launches Mercury 2
0tok/s
Generation speed
0x
Faster than speed-optimized
0k
Context window
Diffusion vs Autoregressive
1Start from noise
2Parallel refine
3Final output
0%
IFBench
0
AIME 2025
Pricing Per Million Tokens
Input0.25
Output0.75
Speed Meets Reasoning
AI NEWS BLITZ
Inception Labs has released Mercury 2, a reasoning model built on diffusion.