Skip to content

Benchmark

Multi-SWE-bench TypeScript Verified

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

Models measured
10
Scores recorded
31
Sources
1
IQ weight
1.33

Score distribution

5.8%
Mean
4.5%
Median
6.2%
Std dev
0.446%
10th percentile
11.6%
90th percentile
0%
Min
0%27.2%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.33 for chat models, measured over 10 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Claude 3.5 SonnetRepoRepair, 20250910_RepoRepair_Claude-3.5-Sonnet(Oct)· Multi-SWE-bench Experiments data docs27.2%
2Gemini 2.5 ProMopenHands, 20250426_MopenHands_Gemini-2.5-Pro· Multi-SWE-bench Experiments data docs22.3%
3Claude 3.5 SonnetMopenHands, 20250329_MopenHands_Claude-3.5-Sonnet(Oct)· Multi-SWE-bench Experiments data docs11.6%
4Gemini 2.5 ProMagentLess, 20250426_MagentLess_Gemini-2.5-Pro· Multi-SWE-bench Experiments data docs11.6%
5Claude Sonnet 3.7MSWE-agent, 20250329_MSWE-agent_Claude-3.7-Sonnet· Multi-SWE-bench Experiments data docs11.2%
6Llama 4 MaverickMagentLess, 20250426_MagentLess_Llama-4-Maverick· Multi-SWE-bench Experiments data docs9.4%
7Gemini 2.5 ProMSWE-agent, 20250426_MSWE-agent_Gemini-2.5-Pro· Multi-SWE-bench Experiments data docs8.9%
8Claude 3.5 SonnetMSWE-agent, 20250329_MSWE-agent_Claude-3.5-Sonnet(Oct)· Multi-SWE-bench Experiments data docs8%
9Llama 4 MaverickMopenHands, 20250426_MopenHands_Llama-4-Maverick· Multi-SWE-bench Experiments data docs8%
10DeepSeek V3MagentLess, 20250329_MagentLess_DeepSeek-V3· Multi-SWE-bench Experiments data docs6.7%
11DeepSeek R1MagentLess, 20250329_MagentLess_DeepSeek-R1· Multi-SWE-bench Experiments data docs6.3%
12DeepSeek R1MSWE-agent, 20250329_MSWE-agent_DeepSeek-R1· Multi-SWE-bench Experiments data docs5.8%
13o1MagentLess, 20250329_MagentLess_OpenAI-o1· Multi-SWE-bench Experiments data docs5.8%
14Claude 3.5 SonnetMagentLess, 20250329_MagentLess_Claude-3.5-Sonnet(Oct)· Multi-SWE-bench Experiments data docs4.9%
15o3 MiniMSWE-agent, 20250329_MSWE-agent_OpenAI-o3-mini-high· Multi-SWE-bench Experiments data docs4.9%
16Llama 4 MaverickMSWE-agent, 20250426_MSWE-agent_Llama-4-Maverick· Multi-SWE-bench Experiments data docs4.5%
17Qwen2.5 72B InstructMagentLess, 20250329_MagentLess_Qwen2.5-72B-Instruct· Multi-SWE-bench Experiments data docs4.5%
18o1MSWE-agent, 20250329_MSWE-agent_OpenAI-o1· Multi-SWE-bench Experiments data docs4%
19Claude Sonnet 3.7MagentLess, 20250329_MagentLess_Claude-3.7-Sonnet· Multi-SWE-bench Experiments data docs3.6%
20DeepSeek V3MSWE-agent, 20250329_MSWE-agent_DeepSeek-V3· Multi-SWE-bench Experiments data docs2.7%
21Claude Sonnet 3.7MopenHands, 20250329_MopenHands_Claude-3.7-Sonnet· Multi-SWE-bench Experiments data docs2.2%
22GPT 4oMagentLess, 20250329_MagentLess_GPT-4o-1120· Multi-SWE-bench Experiments data docs2.2%
23DeepSeek V3MopenHands, 20250329_MopenHands_DeepSeek-V3· Multi-SWE-bench Experiments data docs1.3%
24DeepSeek R1MopenHands, 20250329_MopenHands_DeepSeek-R1· Multi-SWE-bench Experiments data docs0.446%
25GPT 4oMSWE-agent, 20250329_MSWE-agent_GPT-4o-1120· Multi-SWE-bench Experiments data docs0.446%
26o1MopenHands, 20250329_MopenHands_OpenAI-o1· Multi-SWE-bench Experiments data docs0.446%
27o3 MiniMagentLess, 20250329_MagentLess_OpenAI-o3-mini-high· Multi-SWE-bench Experiments data docs0.446%
28o3 MiniMopenHands, 20250329_MopenHands_OpenAI-o3-mini-high· Multi-SWE-bench Experiments data docs0.446%
29GPT 4oMopenHands, 20250329_MopenHands_GPT-4o-1120· Multi-SWE-bench Experiments data docs0%
30Qwen2.5 72B InstructMopenHands, 20250329_MopenHands_Qwen2.5-72B-Instruct· Multi-SWE-bench Experiments data docs0%
31Qwen2.5 72B InstructMSWE-agent, 20250329_MSWE-agent_Qwen2.5-72B-Instruct· Multi-SWE-bench Experiments data docs0%

Data credits: Multi-SWE-bench experiments (multi-swe-bench/experiments), Apache-2.0; resolved-rate fractions converted to percentages (licence)