# Reflection AI

> Reflection AI is a New York-based artificial intelligence lab founded in 2024 by former Google DeepMind researchers Misha Laskin and Ioannis Antonoglou. It is building autonomous coding agents and open frontier language models, with the stated ambition of reaching superintelligence by first solving software engineering. Its flagship product, Asimov, is a code research and comprehension agent for large codebases. After raising $2 billion at an $8 billion valuation in October 2025, the company has positioned itself as an open, Western alternative to closed labs like OpenAI and Anthropic and to Chinese open-weight models like DeepSeek.

- **Founded:** 2024
- **Headquarters:** New York, New York, United States
- **Founders:** Misha Laskin (Co-Founder & CEO), Ioannis Antonoglou (Co-Founder & CTO)
- **Team size:** ~60 employees
- **Products:** Asimov, Open frontier language model, Reinforcement learning post-training stack
- **Notable:** Raised $2 billion at an $8 billion valuation in October 2025, one of the largest early-stage AI rounds to date., Grew valuation roughly 15x in seven months (from ~$545M to $8B)., Founders helped build landmark systems including AlphaGo (Antonoglou) and reward modeling for Gemini (Laskin).

## Products & services

- **Asimov** — An autonomous code research and comprehension agent for large codebases. It ingests not only source code but also documentation, Slack messages, emails and project notes to understand how and why software was built. Uses a multi-agent architecture - many small long-context retriever agents feeding a large short-context reasoning agent - and introduces 'Asimov Memories' to capture team-wide engineering knowledge.
- **Open frontier language model** — A frontier Mixture-of-Experts language model trained on tens of trillions of tokens, planned for release in early 2026 with public model weights. Initially text-based with planned multimodal capabilities.
- **Reinforcement learning post-training stack** — Proprietary scalable RL and post-training techniques that let models plan, debug, test, refactor and iterate over entire software systems rather than predict single lines of code.

## Achievements

- Raised $2 billion at an $8 billion valuation in October 2025, one of the largest early-stage AI rounds to date.
- Grew valuation roughly 15x in seven months (from ~$545M to $8B).
- Founders helped build landmark systems including AlphaGo (Antonoglou) and reward modeling for Gemini (Laskin).
- Launched Asimov in July 2025, with maintainers preferring its answers 60-80% of the time in blind tests.
- Assembled what it describes as the highest-density RL talent of any startup.

## Latest updates

- **2025-03** — Reflection AI launched publicly with $130 million in early-stage funding led by Lightspeed, Sequoia and CRV with NVIDIA participation.
- **2025-07** — Launched Asimov, an autonomous code comprehension agent for large codebases that learns from code, docs, Slack and emails.
- **2025-10** — Raised a $2 billion Series B at an $8 billion valuation, repositioning as an open frontier AI lab challenging DeepSeek and closed labs.
- **2026-01** — Company targeting release of its first open-weights frontier Mixture-of-Experts model trained on tens of trillions of tokens in early 2026.

## Links

- Website: https://reflection.ai
- LinkedIn: https://www.linkedin.com/company/reflectionai/
- Twitter/X: https://x.com/reflection_ai

---

Profile page: https://yespress.io/reflection-ai
Published by YesPress — https://yespress.io
Last updated: 2026-06-13