Gemini 4 Argon on NExT-QA
90.0%Rank #2 of 10 on NExT-QA
- Coverage
- Full (4996/4996)
- Cost per task
- $0.0092
- Median latency
- 4.8s
- Tokens per task
- 4,527 in · 14 out
Run setting: 1 fps.
Tokens and latency
Tokens per call
| Token type | Mean | Median | 90th pct. |
|---|---|---|---|
| Input | 4,527 | 3,787 | 9,195 |
| of which cached | 697 | 0 | 3,100 |
| Output | 14 | 5 | 5 |
| of which reasoning | 9 | 0 | 0 |
Over 10,096 calls that returned an answer, including questions asked again after a failed attempt.
Latency
- Median
- 4.8s
- Mean
- 8.4s
- 90th percentile
- 12.5s
- 99th percentile
- 80.7s
Per call, end to end including file upload and processing.