OpenAI overtakes Anthropic on OpenRouter spending for the first time in 2.5 years
新智元 · wechat · 2026-09-16
OpenRouter data shows that in the week of Sept 7, users' spending on OpenAI models surpassed Anthropic's for the first time in over 30 months — a meaningful reversal, since wallet share is seen as a truer signal than raw token volume.
- Four-model lineup: GPT-6 Astra, launched just two weeks ago, captured 19% of combined spending, beating Claude Opus 5 (16%); priced at $10/M input and $50/M output tokens
- Dual-track strategy: GPT-5.6 Luna undercut Astra's price by 50x while carrying far more volume; GPT-5.6 Sol and Terra each took 10%
- GPT-6 Sol leak: some developers hitting GPT-5.6 Sol got routed to "GPT-6 Sol"; early testers report near-crushing工程质量 vs Opus 5.2 and 6x faster than Astra on an SVG task (3 vs 19 minutes)
- What's next: Altman teased major releases this week ahead of DevDay, where the GPT-6 family is expected to debut; Anthropic is graying out Opus 5.2
(GPT-6 Sol details are unconfirmed leaks.)
More from Models
- LightOnOCR-3 Reads 16th-Century Manuscripts and Ancient Greek Without Training — IgorCarron · 2026-11-02
- 'You Can Feel the Gradient Slopes': Dev Contrasts Benchmaxxed RL Models vs General Ones — theteknosaur · 2026-11-02
- LightOnOCR-3 Launches: Tops FRBench, Adds Layout Structure and Chart Extraction — IgorCarron · 2026-11-02
- NVIDIA's Kumo Is a Family of Four Tabular Foundation Models, Not Just One — pandeyparul · 2026-11-02
- SemiAnalysis: safety classifiers are underrated — Meta's Muse tops cyber benchmark at 35.2% — ThomasScialom · 2026-11-02
- DeepSeek V4.1 Flash 1017 wins over a longtime closed-source-only user: fast, cheap, reliable — _bachrc · 2026-11-02