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.

Eight open biology models, free to use, on hardware we own and operate in Japan. This is the inference infrastructure behind them, and why serving this workload looks almost nothing like serving a language model.

ai& and Tenstorrent launched JapanFold, a drug discovery platform running on ai&'s sovereign infrastructure, powered by Tenstorrent Galaxy™ superclusters. Japanese researchers, pharmaceutical companies, and biotech firms can now access frontier drug discovery models at a fraction of the cost of conventional cloud infrastructure, with every computation running entirely within Japan.

Kimi K3 is being called “the clearest example yet of an open-weight model reaching frontier-level performance.” Moonshot AI’s newly released 2.8-trillion-parameter model scores 57 on the Artificial Analysis Intelligence Index — behind only Claude Fable 5 (60) and GPT-5.6 Sol (59), and ahead of every other model tested, including Claude Opus 4.8.

ai&, a vertically integrated global AI technology company, today launched ai& inference, a high-performance platform that delivers state of the art inference at a fraction of what comparable proprietary inference cost to run today. The platform is built on a heterogeneous compute architecture, combining AMD, NVIDIA, and other silicon architectures under a single optimized inference serving layer. ai& inference is engineered to convert hardware-software co-design into a structural cost advantage that single-architecture providers cannot match.

ai& a vertically integrated global AI technology company, and Plug and Play Japan KK, one of the world’s largest innovation platforms, today announced a collaboration to help enterprises build and deploy AI with greater speed, control, and confidence. The partnership connects Plug and Play's AI Center of Excellence (AI CoE) with ai&'s vertically integrated AI platform, which unifies data centers, heterogeneous compute, and advanced model services into a single stack built from Japan for the global market.

AI has shifted from experimentation to foundational business infrastructure — and the organizations that move fastest from ambition to execution will define the next era. ai& CEO David Bennett and Jon Robottom, CVP and GM of AMD Japan, sit down to discuss what it actually takes to deploy AI at scale: from agent-to-agent architectures and compute economics, to the opportunity in Japan and beyond. The infrastructure to participate fully in the AI era is available today.

ai& Inc. (headquartered in Yokohama, Kanagawa; Co-Founder & CEO: David Bennett; hereinafter “ai&”), a vertically integrated global AI technology company, today announced that it has entered into a partnership with Ryoyo Ryosan, Inc. (headquartered in Chuo-ku, Tokyo; President: Kazuhiko Inaba; hereinafter “Ryoyo Ryosan”), an electronics trading company specializing in semiconductors, AI, IT, and industrial DX.

ai& Inc., a vertically integrated global AI technology company headquartered in Yokohama, Kanagawa, Japan, and Tomorrow Net Co., Ltd. (headquartered in Shinagawa-ku, Tokyo; President & CEO: Lee Chang-Jin) today announced that they have entered into a business alliance to provide generative AI and AI/IT infrastructure solutions for enterprise customers.

ai& today announced that the company is collaborating with GENSHI AI Inc., a medical AI company, to build a secure large language model (LLM) platform for healthcare institutions in Japan. The companies are currently conducting a proof of concept (PoC), with plans for full-scale rollout in fiscal year 2026. Together, they aim to enable advanced AI while keeping sensitive data within the country.