Inception's Mercury 2.5 pushes a diffusion language model past 1,100 tokens a second
Diffusion models refine a whole draft in parallel instead of emitting one token at a time. That is the entire pitch, and the throughput figure is where it shows.
Company
Inception is an organisation working in artificial intelligence, founded in 2024 and based in Palo Alto, United States. This page collects every Aivio News story about Inception, together with its models and recorded benchmark results.
Inception's Mercury 2.5 pushes a diffusion language model past 1,100 tokens a second
Diffusion models refine a whole draft in parallel instead of emitting one token at a time. That is the entire pitch, and the throughput figure is where it shows.
Privacy preferences
We use necessary cookies to run this site. With your consent we also use advertising cookies. You can change your choice at any time. Cookie Policy · Privacy Policy