Inference
Deploy
Company
Contact
Docs
Run inference
Talk to our team
What should I see in Kyoto in autumn?
Start early at Fushimi Inari before the crowds, then walk the maple trails at Tofuku-ji. In the evening, the lantern-lit streets of Gion are at their best.
Run leading open models faster, at lower cost.
Experience lightning-fast video generation from text, photos, or video clips.
Explore protein structures to support your drug discovery research.
Make any of them yours.Shape a model around your language, expertise and workflows.
Explore post-training
Push your code to launch an app, with review apps, TLS, and scaling built in. Hosted in Japan.
We own our data-centers in Japan. With 100MW+ capacity of NVIDIA, AMD, Tenstorrent, and Rebellions chips in the works.
Visit the newsroom
What we learned running H3 on NVIDIA, AMD, and Tenstorrent and how we approached the tradeoffs between speed, cost, and quality.
Blog
Oct 22, 2026
ai& today launched ai& video, an AI video generation service that turns text, photos, and existing video clips into new footage, generating video in as little as 12 seconds on ai&'s own infrastructure in Japan.
News
Graded six times over; all 36,144 attempts released, the 14,234 failures included. Every task carries a measured pass rate per model, not a difficulty label somebody chose, and that measurement picked the RL training data: +3.1 on Terminal-Bench 2.1, and +2.1 across five runs of the same recipe.
Oct 8, 2026
Bring your next idea to life with ai&.
Get started
