Nemotron 3 Nano Omni 30B-A3B on UCF101-AD
31.0%Rank #8 of 9 on UCF101-AD
- Coverage
- Full (4224/4224)
- Cost per task
- —
- Median latency
- 3.2s
- Tokens per task
- 1,345 in · 1,129 out
Run setting: 1 fps.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 1,345 | 1,141 | 1,693 |
| of which cached | 0 | 0 | 0 |
| Output | 1,129 | 616 | 2,175 |
Over 4,224 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 3.2s
- Mean
- 5.7s
- 90th percentile
- 10.5s
- 99th percentile
- 40.0s
We only consider the first question against a video for self hosted models, ignoring prefix caching latency.