OpenRouter Token Consumption Up 224x, Spending 81x Since Early 2026

Token consumption on OpenRouter grew 224 times and spending grew 81 times since early 2026, with the platform crossing a 10 quadrillion token annual run rate and volumes doubling every 10.5 weeks. The last 9 weeks alone exceeded all prior cumulative usage on the platform. OpenAI accounts for 52% of tokens and 34% of spending, while Anthropic captures 66% of spending with 48% of tokens. A Menlo Ventures report found the average company spent 4.7 times its planned 2025 AI budget against a forecast of only 1.9 times, with the median company overshooting more modestly at 2.2 times versus a planned 1.8 times; Menlo Ventures attributed most of the surge to a small group of heavy AI agent users. Fireworks AI reached 40 trillion tokens per day as of mid-July 2026, up from 15 trillion in April and 10 trillion in October 2025. Microsoft has integrated Fireworks AI's inference engine into Azure through Microsoft Foundry, with Fireworks processing over 13 trillion tokens daily at approximately 180,000 requests per second through that integration. Microsoft Foundry provides a single control plane to evaluate, deploy, customize, and operate open models, and the partnership supports custom weight model deployments alongside standard open models.