OpenRouter users spent more on OpenAI's models than on Anthropic's in the week of September 7, the first time that happened since the week of February 26, 2024
OpenRouter users spent more on OpenAI models than on Anthropic models last week. This hasn't happened for more than 2.5 years
Context & Ripple Effects
OpenRouter’s model mix had already been fragmenting: reasoning models accounted for more than half of usage over the prior year, while lower-cost Chinese models overtook US rivals in token consumption on the platform in February. A spend lead between OpenAI and Anthropic is therefore a shift within a more contested routing market, not a measure of all AI demand.
The platform signal also differs from enterprise-newcomer purchasing data: Ramp reported in March that Anthropic captured about 73% of spending among companies buying AI tools for the first time. The contrast makes channel-specific spend data more important for both vendors’ sales narratives.
First-order effects
- OpenAI takes the top spend position among OpenRouter users for the week of September 7, displacing Anthropic from a lead it had held on that platform since February 2024.
- Anthropic loses a visible wallet-share benchmark among multi-model users, even as OpenRouter remains only one purchasing channel.
Second-order effects
- OpenRouter customers gain a fresh platform-level comparison point when allocating workloads between OpenAI and Anthropic, reinforcing procurement decisions based on observed spend and model mix rather than a single vendor’s market-share claim.
- OpenAI and Anthropic must reconcile divergent channel signals: OpenRouter spend has shifted toward OpenAI, while Ramp’s March data showed Anthropic leading first-time AI-tool spending.
Third-order effects
- If spend leadership continues to rotate across routing platforms, model vendors will compete less on a durable single-provider position and more on retaining workloads that buyers can switch among providers.
- Multi-model procurement strengthens buyer power: platform-level usage and spend data can make performance, cost, and availability comparisons more consequential than vendor-wide revenue narratives.
The trend: AI model buying is becoming increasingly channel-specific, with routing platforms exposing faster shifts in workload allocation among competing proprietary and open-weight models.