Legal Research Bench
Results
26 of 41 tracked models have a result.
- 1
Anthropic · Max effort
55.3% - 1
Meta · Max effort
55.3% - 3Gemini 4 ArgonLimited
Google DeepMind · High effort
54.8% - 4
Anthropic · Max effort
50.5% - 5GLM-5.3Open
Z.ai · Max effort
49.0% - 6
Anthropic · Max effort
48.1% - 7
Alibaba
47.6% - 8
SpaceXAI · xHigh effort
47.1% - 8MiMo-V2.6-ProOpen
Xiaomi
47.1% - 10Kimi K3Open
Moonshot AI · Max effort
46.2% - 11Hy4 previewOpen
Tencent
45.2% - 12
Anthropic · Max effort
43.3% - 13
DeepSeek · High effort
41.4% - 13
OpenAI · Max effort
41.4% - 15DeepSeek V4 ProOpen
DeepSeek · Max effort
40.9% - 16
StepFun
40.4% - 17
OpenAI · Max effort
39.4% - 18
Google DeepMind · High effort
38.9% - 19
OpenAI · Max effort
38.5% - 20MiMo-V2.6-FlashOpen
Xiaomi
38.0% - 21
Mistral AI · High effort
31.7% - 22
OpenAI · Max effort
30.3% - 23MiniMax-M3Open
MiniMax
29.8% - 24InklingOpen
Thinking Machines Lab
28.4% - 25Inkling-SmallOpen
Thinking Machines Lab
25.5% - 26
Google DeepMind · High effort
20.7%
15 tracked models without a result
Command A+, Gemini 3.5 Flash-Lite, Gemma 4 31B, GLM-5.3-Flash, gpt-oss-120b, Hy3, Kimi K2.7 Code, MAI-Thinking-1, Mistral Medium 3.5, Mistral Small 4, Muse Glimmer 30B, Nemotron 3 Ultra, Nova 2 Lite, Qwen3.8-27B, Qwen3.8-Flash.
Missing means the evaluator has not published a run, not that the model would score poorly.