OpenRouter raises $113 million at $1.3 billion valuation

OpenRouter, an AI gateway startup that provides software that helps users select and access different AI models, has raised $113 million in Series B funding. The investment round was led by CapitalG, the growth venture fund of Alphabet. While the startup did not disclose its new valuation, The New York Times reported that the post-money valuation is about $1.3 billion. This represents a substantial increase from the company’s previous post-money valuation, which PitchBook estimated at $547 million. That previous valuation was established after OpenRouter raised $40 million in Series A funding in June 2025. The Series A round was led by Andreessen Horowitz and Menlo Ventures, with participation from Sequoia.

The funding comes as AI workloads shift from model training to inference and autonomous agents. Founded in 2023, OpenRouter has experienced rapid adoption as developers and enterprises scale their artificial intelligence operations. The platform’s growth is reflected in its key operational metrics:

• Global user base: The platform now serves 8 million global users.
• Monthly token volume: OpenRouter processes 100 trillion tokens per month.
• Weekly token volume: The platform processes 25 trillion tokens per week.
• Growth rate: This weekly volume represents a 5x increase from the 5 trillion tokens processed per week just six months ago.

This volume expansion highlights how AI workloads are shifting toward inference and autonomous agents. Rather than standardizing on a single model vendor, enterprises are increasingly using AI gateways to maintain flexibility and avoid vendor lock-in. OpenRouter provides access to over 400 models, allowing users to route tasks to different model providers—including Anthropic, Google, OpenAI, xAI, and DeepSeek—depending on the specific requirements of the task. This approach allows companies to treat individual models as swappable engines for specific AI tasks.

Why it matters

OpenRouter’s growth indicates that companies are moving toward a multi-model future rather than standardizing on a single model vendor, with the AI gateway serving as an invisible, swappable engine for AI tasks.