Sakana AI

Sakana AI

Sakana AI is the Tokyo-based frontier AI research lab founded by David Ha (ex-Google Brain) and Llion Jones (co-author of the original Transformer paper). Sakana AI's research focuses on nature-inspired AI — evolutionary model merging, biologically grounded architectures, and automated discovery — rather than scaling existing transformer recipes. Sakana AI raised approximately $200M+ at a $1B+ valuation and is the most-watched frontier lab outside the US-China duopoly.

Production credibility: Founded 2023 by David Ha (ex-Google Brain, ex-Stability) and Llion Jones (Transformer co-author, ex-Google); Tokyo HQ. Approximately $200M+ raised at $1B+ valuation; investors include NEA, Khosla Ventures, Lux Capital, and Japan's NTT. Published the Evolutionary Model Merge paper (2024) and AI Scientist agent (2024).

Key Features

• Evolutionary model merging — combine multiple open-weight LLMs to produce better models without retraining
• AI Scientist — autonomous research agent that proposes hypotheses, runs experiments, writes papers
• Japanese-language foundation model focus alongside multilingual research
• Open-weight model releases including EvoLLM, EvoVLM, and Tinyswallow series
• Headquartered in Tokyo with backing from NTT and Japanese government AI initiatives
• Approximately $200M+ raised at $1B+ valuation across seed and Series A
• Investor base: NEA, Khosla Ventures, Lux Capital, Sony Group, NTT

Ideal Use Case

AI researchers, foundation-model teams, and enterprise R&D groups tracking the next architectural shift in AI — particularly companies that want non-US-aligned frontier research partners or have specific Japanese-language deployment requirements.

How Sakana AI differentiates

Most frontier labs (OpenAI, Anthropic, Google DeepMind) compete on scale — bigger compute, more data, longer training runs. Sakana AI bets on the opposite hypothesis: nature-inspired techniques like evolutionary merging can produce frontier capabilities at a fraction of the compute. The Transformer-author pedigree gives Sakana research credibility; the Tokyo geography makes Sakana the only frontier lab with deep ties to Japanese industry and government. For organizations looking for a non-US-aligned research partner with credible architecture work, Sakana is the default choice.

FAQ

Q: Who founded Sakana AI? A: David Ha (ex-Google Brain, ex-Stability) and Llion Jones (co-author of the original Transformer paper at Google) co-founded Sakana AI in 2023. The company is headquartered in Tokyo.

Q: How much has Sakana AI raised? A: Approximately $200M+ across rounds at a $1B+ valuation. Investors include NEA, Khosla Ventures, Lux Capital, Sony Group, and NTT.

Q: What is evolutionary model merging? A: A Sakana AI research direction that combines multiple open-weight LLMs using evolutionary algorithms — producing models that outperform their parents without additional training compute.

Q: Sakana AI vs OpenAI vs Anthropic? A: OpenAI and Anthropic compete on scaling existing transformer architectures with more compute. Sakana bets on nature-inspired techniques (evolutionary merging, biologically grounded architectures) to achieve frontier capabilities with less compute. Different research philosophy.

Q: Is Sakana AI's work open source? A: Sakana releases multiple open-weight model series including EvoLLM, EvoVLM, and Tinyswallow. Research papers are typically published openly on arXiv.

tl;dr

Sakana AI is the Tokyo frontier lab founded by ex-Google Brain David Ha and Transformer co-author Llion Jones. ~$200M+ raised at $1B+ valuation. Nature-inspired AI research — evolutionary model merging, AI Scientist agent. Backed by NEA, Khosla, Lux, Sony, NTT. The non-US/non-China frontier lab worth watching.

Related

Looking for more options? Browse the AI/ML Models directory or read our best AI models listicle. Sakana AI is also tracked on Crunchbase.