Nemotron 3 Nano Omni 30B-A3B on Perception Test
70.3%Rank #7 of 9 on Perception Test
- Coverage
- Full (454/454)
- Cost per task
- —
- Median latency
- 8.2s
- Tokens per task
- 3,592 in · 3,887 out
Run setting: 1 fps.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 3,592 | 3,535 | 5,178 |
| of which cached | 0 | 0 | 0 |
| Output | 3,887 | 919 | 17,411 |
Over 454 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 8.2s
- Mean
- 26.8s
- 90th percentile
- 97.1s
- 99th percentile
- 111.1s
We only consider the first question against a video for self hosted models, ignoring prefix caching latency.