Benchmarks · Vals AI
Immigration Bench
Measures 262 expert-written questions on US immigration law (seven personas), graded per question rubric; split into all-pass, intake checks and content checks.
- Models measured
- 26
- Top
- 75.44 %Claude Fable 5.1within the standard error of ±1.40
- Version
- 1
- Measured
- 9 November 2026
01
Ranking
At the top: Top 15
Vals AI Legal Research Bench, all-pass [1]. Hatched bars are rejected by the compliance gate.
- 1Claude Fable 5.175.44%
- 2Claude Fable 5.273.98%
- 3Claude Opus 5.572.14%
- 4Wonder Full71.22%
- 5Qwen 3.8 Max70.14%
- 6Wonder69.62%
- 7Claude Sonnet 5.568.96%
- 8Kimi K368.26%
- 9Gemini 4 Argon68.08%
- 10Grok 4.768.05%
- 11GPT-6.1 Sol67.81%
- 12GPT-6 Astra67.69%
- 13MiMo V2.6 Pro67.50%
- 14GLM 5.366.73%
- 15Muse Spark 1.266.54%
Source: Vals AI (Immigration Bench) · 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.
| # Model | Score | Cost/task | Gate | PrudentScore |
|---|---|---|---|---|
| 1Claude Fable 5.1Anthropic | 75.44%± 1.40 | $2.11 | Rejected | no score |
| 2Claude Fable 5.2Anthropic | 73.98%± 1.33 | $0.64 | Rejected | no score |
| 3Claude Opus 5.5Anthropic | 72.14%± 1.45 | $2.16 | Passes | 63 |
| 4Wonder FullMeta | 71.22%± 1.46 | $0.39 | Rejected | no score |
| 5Qwen 3.8 MaxAlibaba | 70.14%± 1.57 | $0.12 | Not yet assessed | no score |
| 6WonderMeta | 69.62%± 1.44 | $0.30 | Rejected | no score |
| 7Claude Sonnet 5.5Anthropic | 68.96%± 1.54 | $1.16 | Passes | 62 |
| 8Kimi K3Moonshot AI | 68.26%± 1.51 | $0.18 | Rejected | no score |
| 9Gemini 4 ArgonGoogle | 68.08%± 1.45 | $0.51 | Passes | 79 |
| 10Grok 4.7xAI (SpaceXAI) | 68.05%± 1.60 | $0.24 | Rejected | no score |
| 11GPT-6.1 SolOpenAI | 67.81%± 1.53 | $0.23 | Passes | 59 |
| 12GPT-6 AstraOpenAI | 67.69%± 1.55 | $1.37 | Passes | 56 |
| 13MiMo V2.6 ProXiaomi | 67.50%± 1.55 | $0.02 | Rejected | no score |
| 14GLM 5.3Z.ai | 66.73%± 1.62 | $0.12 | Rejected | no score |
| 15Muse Spark 1.2Meta | 66.54%± 1.59 | $0.04 | Rejected | no score |
| 16Grok 4.6xAI (SpaceXAI) | 65.98%± 1.53 | $0.10 | Rejected | no score |
| 17DeepSeek V4.1 FlashDeepSeek | 65.64%± 1.64 | $0.05 | Rejected | no score |
| 18Muse Spark 1.3Meta | 65.00%± 1.56 | $0.05 | Rejected | no score |
| 19Muse Spark 1.3 (Max)Meta | 64.93%± 1.56 | $0.06 | Rejected | no score |
| 20Mistral Large 4 (preview)Mistral AI | 63.71%± 1.66 | $0.15 | Rejected | no score |
| 21Gemini 3.8 FlashGoogle | 63.41%± 1.61 | $0.17 | Passes | 62 |
| 22GPT-6 LunaOpenAI | 62.53%± 1.57 | $0.01 | Passes | 54 |
| 23Claude Haiku 5.5Anthropic | 61.79%± 1.64 | $0.09 | Passes | 59 |
| 24DeepSeek V4 Pro (0813)DeepSeek | 61.66%± 1.68 | $0.05 | Rejected | no score |
| 25MiniMax-M3MiniMax | 57.53%± 1.62 | $0.04 | Rejected | no score |
| 26Gemini 3.5 Flash LiteGoogle | 55.66%± 1.70 | $0.02 | Passes | 38 |
Source: Vals AI (Immigration Bench) · 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 9 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
- Legal Research Bench
- Harvey LAB (Vals implementation)
- LegalBench (Vals leaderboard)
- CaseLaw v2
- TaxEval v2
- Tax Agent Bench
- CorpFin v2
- Access to Justice
- Public Benefits Bench
- Vals Legal Index
- Vals Index
- GPQA Diamond (Vals-run)
- MMLU Pro (Vals-run)
- Artificial Analysis Intelligence Index
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.