Mirendil
Category: Foundation Models / LLMs
A frontier AI lab building self-improving AI systems that automate AI research and development itself. Mirendil was founded in 2025. The company is led by Behnam Neyshabur. Based in San Francisco, California, USA. Team size: 11-50. Total funding raised: $200M. Latest round: Seed. Key investors include Andreessen Horowitz (a16z, co-lead), Kleiner Perkins (co-lead), NVIDIA.
- Founded
- 2025
- Headquarters
- San Francisco, California, USA
- Team size
- 11-50
- Total funding
- $200M
Value proposition
Democratizes frontier AI R&D by building AI systems capable of doing meaningful AI research autonomously, making advanced AI-driven science and technology development accessible beyond a handful of large labs.
Products and solutions
Frontier AI research models and an autonomous AI R&D system/lab platform (described as 'a coding agent built for AI research that controls its own GPUs') aimed at accelerating the AI research loop itself — ideation, experimentation, and iteration.
Unique value
Building recursive self-improving AI systems specifically designed to accelerate and automate the AI research lifecycle, rather than just augment human researchers.
Target customer
Scientific research labs, biotech/chemistry/robotics companies, and organizations needing frontier AI R&D capability without building their own frontier AI lab.
Industries served
AI Research/R&D, Scientific Discovery, Drug Discovery, Chemistry, Biology, Robotics (as downstream beneficiaries of democratized frontier AI R&D)
Technology advantage
Founding team combines deep frontier-lab research experience (Anthropic Discovery team, Google DeepMind Gemini/Blueshift reasoning team, xAI/Grok) with a research philosophy centered on experiment provenance and dataset/version tracking as core infrastructure for self-improving systems.
How they differentiate
Focused specifically on automating the AI research loop itself (recursive self-improvement of AI R&D), rather than applying AI within a single science vertical; founding team's direct frontier-lab pedigree (Anthropic/DeepMind/xAI) signals technical credibility at a stage most rivals lack.
Main competitors
Sakana AI (AI Scientist - automated research lifecycle), FutureHouse (nonprofit building AI agents for scientific discovery in biology), other frontier AI research automation labs
Key partnerships
Google Cloud (multiyear $100M+ compute partnership using AI Hypercomputer, announced Aug 2026), backed by NVIDIA (compute/investor relationship)
Major milestones
Founded Dec 2025 after founders left Anthropic, formally announced/launched publicly in 2026 with $200M seed at $1B valuation (one of the largest AI seed rounds to date), signed $100M+ Google Cloud AI Hypercomputer partnership (Aug 2026).
Growth metrics
~20-person founding team at time of seed announcement (per company statements); signed $100M+ multiyear Google Cloud compute partnership within months of founding.
Market positioning
Positioned as one of the best-funded and highest-profile new entrants attempting recursive AI self-improvement, competing at the frontier-lab tier rather than as an applied AI tool vendor.
Geographic focus
Primarily US (San Francisco Bay Area), competing globally against other frontier AI labs including Japan-based Sakana AI.
About Behnam Neyshabur
Ex-Member of Technical Staff & Team Lead (co-led Discovery team / AI Scientist) at Anthropic (2024–2025); previously Senior Staff Research Scientist & Gemini Team Lead (co-led Blueshift reasoning) at Google DeepMind (2019–2024); prior research scientist background in deep learning theory and optimization (SAM co-inventor).
Latest news about Mirendil
More Foundation Models / LLMs companies
Official website: https://mirendil.com
