Skip to content
82.4%Rank #4 of 9 on Perception Test
374 of 454 questions answered correctly.
Coverage
Full (454/454)
Cost per task
$0.0586
Median latency
35.4s
Tokens per task
6,717 in · 2,564 out

Run setting: 1 fps.

Tokens and latency

Tokens per call

Token typeMeanMedian90th pct.
Input6,7176,6859,670
of which cached3,8483,8409,472
of which frames or video6,4336,4089,345
Output2,5647278,101
of which reasoning2,5487108,085

Over 459 calls that returned an answer, including questions asked again after a failed attempt.

Latency

Median
35.4s
Mean
98.1s
90th percentile
311.1s
99th percentile
536.7s

Per call, end to end including file upload and processing.