OpenRouter has raised a $113 million Series B led by CapitalG, giving the AI model gateway a reported $1.3 billion valuation as usage of multi-model infrastructure continues to rise.

The round is notable less for the headline valuation than for what investors say is happening underneath it. Menlo Ventures, an existing OpenRouter backer, said the company is now running at roughly a 1.5 quadrillion-token annual pace and serving more than 8 million developers. OpenRouter's own site describes the platform as an AI gateway with access to more than 400 models across 60-plus providers, and lists 100 trillion monthly tokens.

OpenRouter sits between application developers and model providers such as Anthropic, Google, OpenAI, xAI, DeepSeek, and open-weight model hosts. Instead of integrating each provider separately, teams can route requests through one API, compare cost and latency, and switch models by task.

That position has become more important as AI applications move from single-chatbot deployments toward agents, coding tools, retrieval systems, and production workflows that may need different models for different calls. The funding does not prove that model routing will become a permanent layer in the stack, but it shows investors are willing to back the idea that developers will keep using several model families rather than standardizing on one vendor.

The main open question is durability. Gateways can benefit from fragmentation, but they also depend on provider access, reliability, and enough enterprise controls to justify sitting in the middle of critical AI traffic.