Sakana AI developed ALE-Agent, an AI coding agent that achieved 21st place out of 1,000 participants in a live AtCoder Heuristic Competition on May 18, 2025, competing on NP-hard optimization problems. The agent was built on Gemini 2.5 Pro with domain-specific prompts and inference-time scaling, and demonstrated its capability by iteratively refining solutions through techniques like Poisson distribution approximation and simulated annealing optimization. This represents progress in automating algorithm discovery for real-world optimization problems in logistics, factory planning, and power-grid balancing.
Sakana AI published a paper in Nature Machine Intelligence describing Evolutionary Model Merge, a method that uses evolutionary algorithms to automatically combine existing open-source AI models into new foundation models optimized for specific tasks. The company created three Japanese foundation models (EvoLLM-JP, EvoVLM-JP, and EvoSDXL-JP) where their 7B-parameter Japanese math LLM matched the performance of previous 70B-parameter models without requiring gradient-based training. This approach enables automated foundation model development with minimal compute resources by systematically discovering effective model combinations that human experts might not intuitively discover.
Sakana AI launched its Applied Team in 2025 to deploy cutting-edge generative AI and AI Agent technology into real-world applications, particularly in finance and defense sectors. The team currently has approximately 35 members combining AI engineers, domain experts, and business professionals who work in rapid development cycles, with engineers and business staff collaborating closely to translate research into deployed solutions. This integrated structure of researchers, engineers, and industry specialists working together enables faster iteration and allows AI applications to move from automating existing workflows to supporting higher-level strategic decision-making.
Sakana AI held its first Applied Engineer Open House on August 7, 2025, with approximately 70 attendees in-person and over 200 online, showcasing the company's work on AI implementation in finance and defense sectors. The company focuses strategically on financial and defense domains as core business areas, leveraging its world-class research team to solve Japan's critical challenges through applied AI engineering. Sakana AI seeks to expand beyond research publications by building practical AI applications that deeply integrate with customer operations, maintaining close collaboration between its R&D and business teams through informal knowledge-sharing mechanisms.
Sakana AI presented a paper at GECCO'25 proposing M2N2, a method that uses evolutionary algorithms to merge AI models by automatically determining how to partition and combine them rather than requiring manual definition. The approach evolved an MNIST classifier from random networks and successfully merged a math specialist LLM with an agentic specialist LLM to handle both tasks better than existing methods. This evolutionary model fusion approach enables more flexible combination of specialized models while avoiding catastrophic forgetting seen in traditional fine-tuning.
Sakana AI and Daiwa Securities Group have formed a long-term partnership to jointly develop an AI-powered wealth management consulting platform. The platform will use Sakana AI's proprietary AI models to provide personalized financial services across customer segments, with development and implementation proceeding in phases through an established working group. This marks Sakana AI's first partnership with a securities firm and represents Daiwa Securities' execution of its digital innovation strategy to address changing market conditions and customer needs.
Sakana AI and Daiwa Securities Group announced a partnership to develop a Total Asset Consulting Platform powered by Sakana AI's models for personalized financial services. The companies will establish a working group to develop and implement the system in stages, with the platform designed to serve clients ranging from new investors to high-net-worth individuals. The collaboration combines Daiwa's financial expertise with Sakana AI's AI agent technology to create new digital financial services offerings.
Sakana AI launched its Applied Team in early 2025 to commercialize AI technology developed by its research division, focusing initially on finance and defense sectors in Japan. The team has already begun partnerships with major domestic and international clients and is expanding rapidly with professionals from leading tech companies and industry-specific experts. The Applied Team works closely with the 30+ research staff through informal collaboration mechanisms, with both teams maintaining distinct missions while sharing knowledge and insights.
Sakana AI launched a Business Development Division to deploy AI technology for Japanese companies and government agencies addressing business and social challenges. The company is introducing team members from its Applied Team who work on developing AI agents, software engineering, project management, and product delivery, with specific focus on defense applications and long-horizon business tasks. Team members are applying research outcomes to build scalable products that solve real-world problems while developing specialized AI models for environments without cloud connectivity.
Every AI story that matters,
in your inbox by 8am.
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the
day in two minutes. Follow companies and topics for alerts, or get the
briefing in Slack. Free, no spam, unsubscribe anytime.
Reading TLDRocket needs no cookies, and the readership counts we rely on come from
our own cookieless analytics. Google Analytics is the exception: it sets cookies and
reports to Google, so it stays switched off until you allow it. You can change your
mind any time from “Cookie settings” in the footer.
Strictly necessary
Session security and form protection (tldrocket-session,
XSRF-TOKEN, 2 hours). The site cannot work without them,
so they need no consent.
Always on
Google Analytics 4 (_ga,
_ga_<id>, up to 2 years). Measures which
stories and sections readers use. Google acts as a third-party processor and may
store the data outside the EU. No advertising, no profiling, no data sold.
