Kimi K3 on EgoSchema
78.6%Rank #6 of 9 on EgoSchema
- Coverage
- Full (500/500)
- Cost per task
- $0.159
- Median latency
- 20.1s
- Tokens per task
- 49,542 in · 684 out
Run setting: 1 fps.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 49,542 | 49,551 | 49,718 |
| of which cached | 622 | 0 | 0 |
| of which frames or video | 48,365 | 48,420 | 48,420 |
| Output | 684 | 298 | 1,594 |
| of which reasoning | 666 | 270 | 1,577 |
Over 500 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 20.1s
- Mean
- 33.4s
- 90th percentile
- 60.4s
- 99th percentile
- 188.2s
Per call, end to end including file upload and processing.