Skip to content

Benchmark

ExploitBench

Coding benchmarks compare models on programming tasks. Scores are percentages. Higher is better.

Models measured
17
Scores recorded
25
Sources
2
IQ weight
1.45

Score distribution

63.2%
Mean
70%
Median
23.3%
Std dev
32.2%
10th percentile
100%
90th percentile
25.3%
Min
25.3%100%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.45 for chat models, measured over 16 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1GPT-6 Astra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim100%
2GPT-6 AstraOpenAI launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.100%
3GPT-6.2 Astra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim100%
4GPT-6.2 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim100%
5GPT-5.6 SolOpenAI launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.78.5%
6Claude Fable 5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim78%
7Claude Fable 5Z.ai设置· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.78%
8GPT-5.6 SolZ.ai设置· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.76.5%
9Kimi K3.1 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim74.7%
10DeepSeek V4.1 FlashExploitBench API harness / 5 seeds· Reported by DeepSeek· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.73.5%
11GPT-5.6 Sol· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim73.5%
12GPT-5.6 SolOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.73.5%
13Claude Opus 5OpenAI launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.70%
14GLM 5.3· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim54.4%
15GLM 5.3Z.ai设置· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.54.4%
16GPT-5.6 Terra· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim52.9%
17GPT-5.6 TerraOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.52.9%
18GPT 5.5OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.47.9%
19MiMo V2.6 Pro· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim47.9%
20Claude Opus 4.8OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.40%
21GPT-5.6 Luna· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim33.2%
22GPT-5.6 LunaOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.33.2%
23Kimi K3Z.ai设置· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.32.2%
24Qwen3.8 MaxZ.ai设置· Reported by Z.ai· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.28.8%
25MiMo V2.6 Flash· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim25.3%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Data by LLM Stats