Reflection raises 2B to build frontier open intelligence

Reflection, an AI startup focused on frontier open intelligence — also known as open frontier AI — has announced it has raised $2 billion to advance the development of large-scale, open, and accessible AI systems. The company says it has assembled a team of top researchers and engineers from projects including PaLM, Gemini, AlphaGo, AlphaCode, and ChatGPT, and built its own large-scale LLM and reinforcement learning stack.

This marks an important move in the AI landscape, signaling a major investment in open-source approaches at a time when most leading-edge AI models remain proprietary and closed. This could open opportunities for European developers, startups, and research institutions to collaborate and innovate without restrictions.

Building frontier-scale open models

Reflection’s goal is to ensure that AI’s most advanced capabilities remain open and transparent rather than concentrated in the hands of a few tech giants. “We’re building frontier open intelligence accessible to all,” the company said in its announcement.

The company claims to have developed a “large-scale LLM and reinforcement learning platform capable of training massive Mixture-of-Experts (MoE) models at frontier scale,” technology once thought achievable only within the world’s largest research labs. The system has already been applied to autonomous coding, and the team now plans to expand its approach to general agentic reasoning — a key step toward building more capable and adaptive AI systems.

Reflection’s funding round includes investors such as NVIDIA, DST, Lightspeed, B Capital, Sequoia, CRV, and others, and supports sustainable open development. The company says its commercial model aligns with its open intelligence strategy, ensuring the release of competitive open models that combine large-scale pretraining with advanced reinforcement learning techniques.

Openness and safety

Reflection argues that openness enhances safety rather than undermines it. By making its models and methods transparent, the company believes it can enable the wider research community to participate in safety evaluation and risk mitigation. “We believe the answer to AI safety is not ‘security through obscurity’ but rigorous science conducted in the open,” the team stated.

At the same time, the company acknowledges the challenges of open access to powerful AI systems. It is investing in evaluation frameworks, security research, and responsible deployment standards to minimize misuse and ensure safe releases.

A call to collaboration

Reflection frames its mission as both urgent and collaborative. “There is a window of opportunity today to build frontier open intelligence, but it is closing and this may be the last,” the announcement said. The company is actively recruiting researchers, engineers, and partners who share its vision of open, sustainable AI progress.

With its $2 billion war chest and an elite AI team, Reflection aims to position itself as a counterweight to closed AI labs — potentially reshaping the balance of innovation and access in the global AI ecosystem.