GPT-6 Astra on EgoSchema
83.4%Rank #2 of 9 on EgoSchema
- Coverage
- Full (500/500)
- Cost per task
- $0.475
- Median latency
- 8.7s
- Tokens per task
- 47,413 in · 19 out
Run setting: 1 fps.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 47,413 | 47,844 | 47,997 |
| of which cached | 286 | 0 | 0 |
| Output | 19 | 4 | 51 |
| of which reasoning | 13 | 0 | 41 |
Over 500 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 8.7s
- Mean
- 8.8s
- 90th percentile
- 12.2s
- 99th percentile
- 17.1s
Per call, end to end including file upload and processing.