Skip to content

Benchmarks · Vals AI

Access to Justice

Measures to what extent AI agents can guide an unrepresented client through a real legal problem.

Models measured
23
Top
76.37 %Qwen 3.8 Maxwithin the standard error of ±0.68
Version
1
Measured
10 November 2026
Go to the sourceThis benchmark does not count towards the PrudentScore.

01

Ranking

At the top: Top 15

Vals AI Legal Research Bench, all-pass [1]. Hatched bars are rejected by the compliance gate.

  1. 1Qwen 3.8 Max76.37%
  2. 2Claude Sonnet 5.574.94%
  3. 3Wonder Full74.68%
  4. 4Claude Opus 5.573.77%
  5. 5GLM 5.373.70%
  6. 6Grok 4.773.69%
  7. 7MiniMax-M3.1-Flash73.31%
  8. 8DeepSeek V4.1 Flash 101773.11%
  9. 9Wonder72.96%
  10. 10DeepSeek V4.1 Flash72.86%
  11. 11Muse Spark 1.372.18%
  12. 12Muse Spark 1.3 (Max)72.15%
  13. 13Grok 4.671.34%
  14. 14GLM 5.3 Flash71.11%
  15. 15MiMo V2.6 Flash70.75%

Source: Vals AI (Access to Justice) · Licence: CC BY 4.0 · Adaptation: Prudai selected the models, translated the explanations, added the gate verdict and the PrudentScore and rounded figures to two decimals.

Passes the compliance gateRejected by the compliance gate
Access to Justice 1
# ModelScoreCost/taskGatePrudentScore
1Qwen 3.8 MaxAlibaba76.37%± 0.68$0.61Not yet assessedno score
2Claude Sonnet 5.5Anthropic74.94%± 0.70$2.37Passes62
3Wonder FullMeta74.68%± 0.70$1.24Rejectedno score
4Claude Opus 5.5Anthropic73.77%± 0.71$3.38Passes63
5GLM 5.3Z.ai73.70%± 0.71$0.46Rejectedno score
6Grok 4.7xAI (SpaceXAI)73.69%± 0.71$0.82Rejectedno score
7MiniMax-M3.1-FlashMiniMax73.31%± 0.71$0.06Rejectedno score
8DeepSeek V4.1 Flash 1017DeepSeek73.11%± 0.71$0.09Rejectedno score
9WonderMeta72.96%± 0.72$1.10Rejectedno score
10DeepSeek V4.1 FlashDeepSeek72.86%± 0.72$0.11Rejectedno score
11Muse Spark 1.3Meta72.18%± 0.72$0.18Rejectedno score
12Muse Spark 1.3 (Max)Meta72.15%± 0.72$0.22Rejectedno score
13Grok 4.6xAI (SpaceXAI)71.34%± 0.73$0.71Rejectedno score
14GLM 5.3 FlashZ.ai71.11%± 0.73$0.02Rejectedno score
15MiMo V2.6 FlashXiaomi70.75%± 0.73$0.02Rejectedno score
16MiMo V2.6 ProXiaomi69.96%± 0.74$0.05Rejectedno score
17MiniMax-M3MiniMax69.77%± 0.74$0.11Rejectedno score
18Kimi K3.1 ProMoonshot AI69.47%± 0.74$0.96Rejectedno score
19Gemini 4 ArgonGoogle68.48%± 0.75$0.83Passes79
20Kimi K3Moonshot AI68.17%± 0.75$0.45Rejectedno score
21Gemini 3.8 FlashGoogle67.06%± 0.76$0.32Passes62
22GPT-6.1 SolOpenAI65.84%± 0.76$0.69Passes59
23GPT-5.6 LunaOpenAI65.56%± 0.77$0.09Passes56

Source: Vals AI (Access to Justice) · Licence: CC BY 4.0 · Adaptation: Prudai selected the models, translated the explanations, added the gate verdict and the PrudentScore and rounded figures to two decimals.

Vals AI · Version 1 · Measured 10 November 2026 · Go to the source. The PrudentScore is a composition by Prudai B.V. based on publicly published figures and does not represent the view or endorsement of Vals AI, Harvey or Artificial Analysis. Figures from Artificial Analysis appear in running text only, with attribution and a link, as that party's terms require [9].

Other benchmarks

Claude and Anthropic, GPT and OpenAI, Gemini and Google, Grok and xAI, Kimi and Moonshot AI, Muse and Meta, Mistral, DeepSeek, GLM and Z.ai, Harvey, Vals AI and Artificial Analysis are trademarks of their respective owners. They are named for identification only and this implies no partnership, sponsorship or endorsement.