OpenRouter data: token usage exploding, some open-weight models see 10x spend since January
AccBalanced · x · 2026-09-30
OpenRouter's first data newsletter by Peter Walker highlights:
- Token usage is exploding; Feb 6, 2026 may have been the last day humans consumed more tokens than agents.
- The frontier race is narrowing, though closed models still lead in spend share.
- Some open-weight labs now see 10x the spend on their models vs January.
- Bonus data points: GLM-5.3-Flash (Ox Alpha) processed over 20 trillion tokens in 6 days, the biggest model ever on the platform; GPT 5.6 Terra/Luna discounts drove a 13.8x token surge; top paid use cases are agent workflow execution, code generation, and bulk classification.
More from Infra
- All PyTorch Conference China 2026 sessions now available on YouTube — PyTorch · 2026-09-30
- Jensen Huang on AI safety: contain agents, monitor them in real time with BlueField chip — XFreeze · 2026-09-30
- NVIDIA Vera CPU lands at Daytona office after running agentic workloads in testing — IanAndrewsDC · 2026-09-30
- Kafgres embeds a Rust Kafka broker inside Postgres, clients can't tell the difference — JiliJeanlouis · 2026-09-30
- Starlink Mobile goes live in Ecuador, bringing direct-to-cell coverage to the Amazon and Galápagos — elonmusk · 2026-09-30
- Reserve unlocked GPU capacity for startups, not just big logos — AccBalanced · 2026-09-30