Gemini 4 Argon on Perception Test
89.7%Rank #2 of 10 on Perception Test
- Coverage
- Full (454/454)
- Cost per task
- $0.0055
- Median latency
- 6.6s
- Tokens per task
- 2,540 in · 42 out
Run setting: 1 fps.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 2,540 | 2,416 | 3,782 |
| of which cached | 171 | 0 | 1,006 |
| Output | 42 | 5 | 111 |
| of which reasoning | 37 | 0 | 106 |
Over 908 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 6.6s
- Mean
- 9.5s
- 90th percentile
- 17.5s
- 99th percentile
- 52.7s
Per call, end to end including file upload and processing.