Human Frontier
—
Human-calibrated frontier signal, backed by Arena-style preference evidence and separate from raw AgMoBench benchmark composite scores.
Blended Price
Free/M
Input Price
Free/M
Output Price
Free/M
Speed
—
TTFT
—
Axes
Filters
Providers
vcontinuous-leaderboard · tasks-created:2025-01-01:2025-09-01 · gemma-3-27b-it · harness swe-rebench-react@text · 5 attempts · pass@1 · swe-rebench:standardized:128k-or-model-limit · provider gemma-3-27b-it__text · record gemma-3-27b-it__text:1735689600000:1756684800000
{
"passAt5": 11.358574610244988,
"developer": "google",
"instanceType": "model",
"contextPolicy": "128k unless the model supports less",
"attemptsPerTask": 5,
"totalTokenUsage": 0,
"modelReleaseDate": "2025-03-10",
"scoreStandardError": 0.4420809185140136,
"cachedTokenPercentage": null,
"modelReleaseTimestamp": 1741564800000,
"standardContextTokens": 128000,
"averageInstanceCostUsd": 0,
"potentialContamination": true,
"aggregateTaskRangeTimestamp": {
"to": 1756684800000,
"from": 1735689600000
},
"publishedTaskRangeTimestamp": {
"to": 1756684800000,
"from": 1735689600000
}
}Evidence observed 2025-09-01