Vals AI Access to Justice: leaderboard
Vals A2J v1:252private legal-access scenarios withGPT-5simulated client andDeciderv1.2judge; pooled expert-rubric criterion pass rate. Native model effort and versioned complete roster retained.
Metric: Rubric Criterion Pass Rate (%). Source: www.vals.ai. Saturation forecast: Not forecast. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.8 Max | 76.37 | #55 |
| 2 | Claude Sonnet 5.5 (Max) | 74.94 | #18 (Claude Sonnet 5.5) |
| 3 | Muse Wonder Full (Max) | 74.68 | #16 (Muse Wonder Full) |
| 4 | Claude Opus 5.5 (Max) | 73.77 | #5 (Claude Opus 5.5) |
| 5 | GLM-5.3 (Max) | 73.7 | #62 (GLM-5.3) |
| 6 | Grok 4.7 (xHigh) | 73.69 | #43 (Grok 4.7) |
| 7 | MiniMax-M3.1-Flash (Max) | 73.31 | #104 (MiniMax-M3.1-Flash) |
| 8 | DeepSeek V4.1 Flash (1017) (Max) | 73.11 | #38 (DeepSeek V4.1 Flash (1017)) |
| 9 | Muse Wonder (Max) | 72.96 | #27 (Muse Wonder) |
| 10 | DeepSeek V4.1 Flash (Max) | 72.86 | #58 (DeepSeek V4.1 Flash) |
| 11 | Muse Spark 1.3 (xHigh) | 72.18 | #42 (Muse Spark 1.3) |
| 12 | Muse Spark 1.3 (Max) | 72.15 | #42 (Muse Spark 1.3) |
| 13 | Grok 4.6 (High) | 71.34 | #44 (Grok 4.6) |
| 14 | GLM-5.3 Flash (Max) | 71.11 | #119 (GLM-5.3 Flash) |
| 15 | MiMo-V2.6-Flash | 70.75 | #150 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #6 GPT-6 Astra.
Interactive version: theaggregate.ai/benchmark?slug=vals-ai-access-to-justice · How It Works · Data refreshed daily, snapshot 2026-11-12.