Qwen3.5-35B-A3B on Perception Test
66.7%Rank #9 of 9 on Perception Test
- Coverage
- Full (454/454)
- Cost per task
- —
- Median latency
- 11.4s
- Tokens per task
- 2,536 in · 3,132 out
Run setting: 1 fps.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 2,536 | 2,502 | 3,641 |
| of which cached | 0 | 0 | 0 |
| Output | 3,132 | 1,354 | 8,738 |
Over 454 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 11.4s
- Mean
- 23.7s
- 90th percentile
- 61.7s
- 99th percentile
- 125.4s
We only consider the first question against a video for self hosted models, ignoring prefix caching latency.