GPT-6 Astra on Perception Test
92.5%Rank #1 of 9 on Perception Test
- Coverage
- Full (454/454)
- Cost per task
- $0.0629
- Median latency
- 4.6s
- Tokens per task
- 6,118 in · 35 out
Run setting: 1 fps.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 6,118 | 6,110 | 8,814 |
| of which cached | 2,903 | 0 | 8,704 |
| Output | 35 | 19 | 87 |
| of which reasoning | 28 | 9 | 77 |
Over 454 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 4.6s
- Mean
- 5.0s
- 90th percentile
- 7.6s
- 99th percentile
- 14.1s
Per call, end to end including file upload and processing.