OpenRouter data: Top models by spend per task
airesearch12 · x · 2026-08-15
Peter Walker shared data from OpenRouter over the past 30 days, showing the top models by spend for various tasks.
- Key Finding: Rankings by spend are relatively concentrated, with a few models dominating the budget (the orange bars in the chart).
- Comparison: Ranking by token usage reveals a much wider variety of model authors and specific models, highlighting a discrepancy between spend and usage volume.
- This suggests that while high-end models might be used less frequently than some smaller ones, their higher pricing or use in complex tasks captures a larger share of revenue.
More from Companies & People
- Rumors: Anthropic's Next Model Marginal, GPT-6 Seen as Major Competitor — robleclerc · 2026-08-15
- Debate sparked over Anthropic's stance on word choice in writing — HamelHusain · 2026-08-15
- AI Agent production boundaries: What actions still require human approval? — JuniorLeg6988 · 2026-08-15
- NYC rolling out agentic assistance to select subway stations — emax · 2026-08-15
- Chinese AI startup ModelBest begins pre-IPO tutoring for mainland listing — ThereWas · 2026-08-15
- Model marketing hype persists as OpenAI and Anthropic revenues soar — bindureddy · 2026-08-15