Inception Releases Diffusion Language Model Mercury 3
Understand the story
On October 21, 2026, Inception released the diffusion language model Mercury 3. According to The Decoder, the company claims Mercury 3 leads GPT-6 Luna and Claude Haiku 5.5 on 8 of 12 benchmarks. In terms of generation speed, Mercury 3 reaches 1,068 tokens/s, more than 4x Claude Haiku 5.5 (240 tokens/s) and about 8x GPT-6 Luna (130 tokens/s). These figures come only from Inception itself; no third-party verification has been reported yet.
AI overview of reporting · Updated 2 h前
Reporting timeline
Explore different aspects of the story through its reporting.
- The DecoderInception's Mercury 3 diffusion model claims to beat GPT-6 Luna and Claude Haiku 5.5 at a fraction of the wait
Inception has released Mercury 3, a diffusion language model, claiming it leads GPT-6 Luna and Claude Haiku 5.5 on 8 of 12 benchmarks, with generation speeds of 1,068 tokens/s—roughly 4-8 times faster than Claude Haiku 5.5 (240) and GPT-6 Luna (130).
Story heat trend
Not enough continuous observations to show a trend yet.