Skip to content

Benchmark

Arena Agent Steerability

Agentic benchmarks measure how well models carry out tasks that take several steps. Higher scores are better.

Models measured
62
Scores recorded
68
Sources
1
IQ weight
2.12

Score distribution

0
Mean
0.008
Median
0.118
Std dev
-0.147
10th percentile
0.14
90th percentile
-0.298
Min
-0.2980.245
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.12 for chat models, measured over 62 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

Showing 51–68 of 68 results.

Data credits: LMArena, CC BY 4.0 (licence)