Qwen3.8-Max on EgoSchema
85.2%Rank #1 of 9 on EgoSchema
- Coverage
- 99.4% (497/500)
- Cost per task
- $0.0496
- Median latency
- 36.0s
- Tokens per task
- 16,849 in · 2,645 out
Run setting: 1 fps native video.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 16,849 | 15,902 | 20,826 |
| of which cached | 1,303 | 0 | 0 |
| of which frames or video | 16,633 | 15,695 | 20,645 |
| Output | 2,645 | 1,230 | 6,995 |
Over 499 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 36.0s
- Mean
- 67.9s
- 90th percentile
- 159.6s
- 99th percentile
- 271.8s
Per call, end to end including file upload and processing.