Skip to content

Benchmark

ALE-Bench

Coding benchmarks compare models on programming tasks.

Models measured
128
Scores recorded
142
Sources
1
IQ weight
2.25

Score distribution

1,065.1
Mean
988.2
Median
593.5
Std dev
340.8
10th percentile
1,932.8
90th percentile
137.8
Min
137.82,951.3
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.25 for chat models, measured over 128 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model
101Claude Opus 4.116k· Epoch AI Benchmarking Hub data674.8
102Qwen3.6 Plus· Epoch AI Benchmarking Hub data670.2
103Gemini 2.5 Flash· Epoch AI Benchmarking Hub data661.9
104Claude Sonnet 432k· Epoch AI Benchmarking Hub data655.4
105Claude Haiku 4.532k· Epoch AI Benchmarking Hub data653.5
106MiniMax M3· Epoch AI Benchmarking Hub data640
107MiniMax M2.1· Epoch AI Benchmarking Hub data623.8
108Qwen3.5 397B A17B· Epoch AI Benchmarking Hub data621.9
109MiniMax M2.5· Epoch AI Benchmarking Hub data618.2
110Qwen3.5 Plus· Epoch AI Benchmarking Hub data608.9
111MiniMax M2.7· Epoch AI Benchmarking Hub data599.3
112Kimi K2 Thinking· Epoch AI Benchmarking Hub data597.5
113Grok Code Fast 1· Epoch AI Benchmarking Hub data587.7
114GPT OSS 120B· Epoch AI Benchmarking Hub data575.6
115GPT OSS 20b· Epoch AI Benchmarking Hub data566.1
116GPT 4.1· Epoch AI Benchmarking Hub data558.1
117DeepSeek V4 Pronone· Epoch AI Benchmarking Hub data521.7
118MiMo V2.5· Epoch AI Benchmarking Hub data514
119Mistral Small2603· Epoch AI Benchmarking Hub data497.6
120Qwen3 Coder· Epoch AI Benchmarking Hub data461.5
121Qwen3 Coder Plus· Epoch AI Benchmarking Hub data456.5
122Ring 2.6 1T· Epoch AI Benchmarking Hub data432.6
123GLM 4.7· Epoch AI Benchmarking Hub data399.5
124Grok 4.1 Fastreasoning· Epoch AI Benchmarking Hub data394.9
125Qwen3 Max· Epoch AI Benchmarking Hub data370.5
126Qwen3.5 27B· Epoch AI Benchmarking Hub data349.5
127GLM 4.5· Epoch AI Benchmarking Hub data344.8
128GLM 4.6· Epoch AI Benchmarking Hub data340.8
129Qwen3.6 Flash· Epoch AI Benchmarking Hub data326.4
130Gemini 2.5 Flash-Litethinking· Epoch AI Benchmarking Hub data325.9
131DeepSeek V4 Flashnone· Epoch AI Benchmarking Hub data325
132GLM 5.3 Flashhigh· Epoch AI Benchmarking Hub data303.6
133Mercury 2.5high· Epoch AI Benchmarking Hub data301.7
134Kimi K2 0905· Epoch AI Benchmarking Hub data267.1
135Qwen3.5 Flash· Epoch AI Benchmarking Hub data265.9
136Mistral Large 3· Epoch AI Benchmarking Hub data264.7
137Nova Litev1· Epoch AI Benchmarking Hub data236.3
138Qwen3.5 35B A3B· Epoch AI Benchmarking Hub data221.8
139Nemotron 3 Super· Epoch AI Benchmarking Hub data213.9
140Mistral Medium 3.1· Epoch AI Benchmarking Hub data210.2
141Llama 4 Maverick· Epoch AI Benchmarking Hub data173
142Codestral 25.012508· Epoch AI Benchmarking Hub data137.8

Showing 101–142 of 142 results.

Tokens the model generated per task, as measured by the source. Comparable within this benchmark only.

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)