Founded in 2023, OpenRouter operates as an AI model gateway — a layer that sits between applications and the growing universe of AI models, allowing enterprises and developers to route requests to whichever model best fits the task at hand. Rather than committing to a single provider, users can optimize for cost, reasoning performance, or speed depending on the job. The platform currently provides access to more than 400 models from providers including OpenAI, Anthropic, Google, xAI, and DeepSeek.
The company says it now counts 8 million global users and processes roughly 100 trillion tokens per month — about 25 trillion per week. That represents a fivefold increase from the 5 trillion tokens per week it was handling just six months ago.
The growth reflects a broader structural shift in how companies are building with AI. A year ago, much of the industry's attention was on model training. More recently, the focus has moved to inference and, now, autonomous agents. In that environment, the ability to swap models in and out — choosing the right one for each step of an agentic workflow — has become increasingly valuable.
The rise of OpenRouter also points to a growing resistance to vendor lock-in. Enterprises that once standardized on a single SaaS provider appear reluctant to do the same with AI models. Rather than betting on one company's model becoming the definitive standard, companies are building on top of interchangeable infrastructure — and OpenRouter is positioning itself as that infrastructure layer.
CapitalG's decision to lead the round is notable given that Google's own Gemini models compete directly with several of the models OpenRouter routes traffic to. It signals that Alphabet views the multi-model gateway market as significant enough to back independently, regardless of where individual requests ultimately land.
This analysis is based on reporting from TechCrunch.
Image courtesy of Micheile Henderson and Unsplash.
This article was generated with AI assistance and reviewed for accuracy and quality.
