Skip to content

Benchmark

Arena ELO Webdev - Webdev

Coding benchmarks compare models on programming tasks. Higher scores are better.

Models measured
139
Scores recorded
159
Sources
1
IQ weight
1.96

Score distribution

1,484.4
Mean
1,484.4
Median
157.2
Std dev
1,263.4
10th percentile
1,683
90th percentile
1,072
Min
1,0721,823.6
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.96 for chat models, measured over 139 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
151Mistral Large 3· LMArena data1,232.6
152Gemini 2.5 Pro· LMArena data1,226.6
153Grok 4.1Thinking· LMArena data1,216
154Devstral 2 123b Instruct 2512· LMArena data1,195.5
155Granite 4.1 8B· LMArena data1,192.8
156Mercury 2· LMArena data1,170.6
157Grok Code Fast 1· LMArena data1,165.9
158Grok 4 FastReasoning· LMArena data1,159.1
159Devstral Medium2507· LMArena data1,072

Showing 151–159 of 159 results.

Data credits: LMArena, CC BY 4.0 (licence)