Performance
Latency, throughput, recent traffic, and availability from real calls — not eval scores.
| Model | Latency (p50) | Throughput | 7d volume | 7d uptime | Detail |
|---|---|---|---|---|---|
| GPT-5.4GPT | 883ms | 350tok/s | 1 | 100.00% | |
| GPT-5.5GPT | 1,989ms | 158tok/s | 2 | 100.00% | |
| Claude Sonnet 4.6Claude | 3,046ms | 11tok/s | 1 | 100.00% | |
| Claude Opus 4.8Claude | 4,674ms | 3tok/s | 2 | 100.00% | |
| GPT Image 2GPT Image 2 | 26,556ms | 9tok/s | 3 | 100.00% | |
| Claude Haiku 4.5Claude | needs metrics | — | — | 100.00% | |
| GPT-5.3 CodexGPT | needs metrics | — | — | 100.00% | |
| GPT-5.4 MiniGPT | needs metrics | — | — | 100.00% |