Compare models
Up to four models side by side, measured the way they run on live video.
| 2 of 4 models | |||
|---|---|---|---|
| Intelligence Index | best in row,86 | 85 | |
| OCR & Text | 95 | 95 | |
| Document & Chart | best in row,87 | 85 | |
| Scene & Spatial | 84 | best in row,86 | |
| Video QA | best in row,86 | 83 | |
| Grounding & Detection | 76 | 76 | |
| Structured Extraction | 84 | best in row,85 | |
| Output Speed | 55 tok/s-36 tok/s | best in row,91 tok/s | |
| TTFT | 207 ms+70 ms | best in row,138 ms | |
| End-to-End | 12 s+3.0 s | best in row,9.3 s | |
| Real-time (<200 ms) | No | No | |
| Blended $/1M | $6.93+$0.25 | best in row,$6.68 | |
| Cost / Task | $5.11 | best in row,$4.93 | |
| Context | 400k | 400k | |
| Params | – | – | |
| Open? | Closed | Closed | |
| License | Proprietary | Proprietary |
- Claude Opus 4.6
- GPT-5.4