Vals CaseLaw v2 (CaseLaw v2)
We show this table for reference; we do not rank on it.
Vals AI private question-answer benchmark over Canadian court cases.
Evaluation results published by Vals AI. Published leaderboard aggregates and evaluation settings only. Dataset contents are not included.
CaseLaw v2 score on CaseLaw v2 — May 4, 2026
We mirror the published caselaw v2 score view for CaseLaw v2. Grok 4.3 leads the public snapshot at 79.31%, followed by GPT-5.1 (73.42%) and GPT-4.1 (69.88%). We do not use these results to rank models overall.
Grok 4.3
xAI
GPT-5.1
OpenAI
high reasoning
GPT-4.1
OpenAI
high reasoning
54 modelsKnowledgeCurrentDisplay onlyUpdated May 4, 2026
CaseLaw v2 score table (54 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Grok 4.3xAI | 79.31% | Not reported | Closed |
| 2 | GPT-5.1OpenAIhigh reasoning | 73.42% | Not reported | Closed |
| 3 | GPT-4.1OpenAIhigh reasoning | 69.88% | Not reported | Closed |
| 4 | GPT-5 miniOpenAIhigh reasoning | 68.49% | Not reported | Closed |
| 5 | Claude Opus 4.7Anthropic | 68.38% | Not reported | Closed |
| 6 | GPT-5OpenAIhigh reasoning | 66.45% | Not reported | Unknown |
| 7 | GPT-5.5OpenAIxhigh reasoning | 66.24% | Not reported | Closed |
| 8 | GPT-5.2OpenAIxhigh reasoning | 66.02% | Not reported | Closed |
| 9 | Grok 4 0709xAI | 65.81% | Not reported | Unknown |
| 10 | Grok 4 Fast (Reasoning)xAI | 65.70% | Not reported | Closed |
| 11 | Kimi K2 ThinkingMoonshot AI | 65.70% | Not reported | Unknown |
| 12 | Gemini 3.1 Pro PreviewGooglehigh reasoning | 64.84% | Not reported | Unknown |
| 13 | Command A 03 2025Cohere | 64.52% | Not reported | Unknown |
| 14 | Claude Sonnet 4.6Anthropic | 63.99% | Not reported | Closed |
| 15 | Gemini 2.5 ProGoogle | 63.88% | Not reported | Closed |
| 16 | GPT-5.4OpenAIxhigh reasoning | 63.77% | Not reported | Closed |
| 17 | Muse SparkMeta | 63.13% | Not reported | Closed |
| 18 | Claude Opus 4.5 ThinkingAnthropic | 62.59% | Not reported | Closed |
| 19 | Claude Sonnet 4.5 ThinkingAnthropic | 62.16% | Not reported | Closed |
| 20 | Claude Opus 4.6 (Adaptive)Anthropic | 62.06% | Not reported | Closed |
| 21 | Mistral Large 2512Mistral AI | 61.41% | Not reported | Unknown |
| 22 | Kimi K2.6Moonshot AI | 61.20% | Not reported | Open |
| 23 | MiniMax M2.7MiniMax | 60.88% | Not reported | Open |
| 24 | Grok 4.1 Fast (Reasoning)xAI | 60.45% | Not reported | Closed |
| 25 | GPT-4oOpenAIhigh reasoning | 59.70% | Not reported | Closed |
| 26 | Qwen3.5 Plus ThinkingAlibaba | 59.70% | Not reported | Unknown |
| 27 | DeepSeek V4 Pro 0813DeepSeekmax reasoning | 59.38% | Not reported | Open |
| 28 | Kimi K2.5 ThinkingMoonshot AI | 58.73% | Not reported | Unknown |
| 29 | Trinity-Large-ThinkingArcee AI | 57.88% | Not reported | Open |
| 30 | Claude Haiku 4.5 ThinkingAnthropic | 56.48% | Not reported | Closed |
| 31 | Qwen3.5 FlashAlibaba | 55.95% | Not reported | Closed |
| 32 | MiniMax M2.1MiniMax | 55.84% | Not reported | Unknown |
| 33 | Gemini 3 Flash PreviewGooglehigh reasoning | 55.84% | Not reported | Unknown |
| 34 | DeepSeek V3p2 ThinkingFireworks AIhigh reasoning | 55.41% | Not reported | Unknown |
| 35 | Gemini 3.1 Flash Lite PreviewGooglehigh reasoning | 54.98% | Not reported | Unknown |
| 36 | Qwen3 MaxAlibaba | 54.98% | Not reported | Closed |
| 37 | GLM-4.7Z.AI | 54.88% | Not reported | Open |
| 38 | Grok 4.20 0309 ReasoningxAI | 54.45% | Not reported | Unknown |
| 39 | DeepSeek V3p1Fireworks AI | 53.91% | Not reported | Unknown |
| 40 | MiniMax M2.5MiniMax | 53.48% | Not reported | Closed |
| 41 | Qwen3.6-27BAlibaba | 53.16% | Not reported | Open |
| 42 | Gemini 3 Pro PreviewGooglehigh reasoning | 53.05% | Not reported | Unknown |
| 43 | Gemma 4 31b ItGooglehigh reasoning | 52.63% | Not reported | Unknown |
| 44 | GPT-5 nanoOpenAIhigh reasoning | 52.63% | Not reported | Closed |
| 45 | GLM 5 ThinkingZhipu AI | 52.52% | Not reported | Unknown |
| 46 | GPT-5.4 nanoOpenAIhigh reasoning | 51.88% | Not reported | Closed |
| 47 | GPT-5.4 miniOpenAIxhigh reasoning | 51.66% | Not reported | Closed |
| 48 | GLM-5.1Z.AI | 51.55% | Not reported | Open |
| 49 | Qwen3.6 PlusAlibaba | 51.45% | Not reported | Closed |
| 50 | GPT-OSS 120BOpenAI | 48.77% | Not reported | Open |
| 51 | Qwen 3.6 Max (preview)Alibaba | 47.91% | Not reported | Closed |
| 52 | Qwen3 MaxAlibaba | 47.48% | Not reported | Closed |
| 53 | Mistral Medium 3.5Mistral AIhigh reasoning | 44.16% | Not reported | Unknown |
| 54 | GPT-OSS 20BOpenAI | 43.84% | Not reported | Open |
How CaseLaw v2 is shown here
BenchLM mirrors the public Vals AI CaseLaw v2 leaderboard captured from https://www.vals.ai/benchmarks/case_law_v2 and updated by Vals on May 4, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.
CaseLaw v2 is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.
Snapshot
The published CaseLaw v2 snapshot places Grok 4.3 first at 79.31%. The third row is 9.43 points behind. The broader top-10 range is 13.61 points, so the table still separates the published systems.
54 models have been evaluated on CaseLaw v2. The benchmark falls in the Knowledge category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. CaseLaw v2 is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About CaseLaw v2
Year
2026
Tasks
Canadian case-law question answering
Format
Accuracy score
Difficulty
Professional legal retrieval and reasoning
Vals marks CaseLaw v2 as archived. BenchLM mirrors the public leaderboard as display-only historical legal-domain context.
Freshness and provenance
Version
CaseLaw v2 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does CaseLaw v2 measure?
Vals AI private question-answer benchmark over Canadian court cases.
Which model leads the published CaseLaw v2 snapshot?
Grok 4.3 currently leads the published CaseLaw v2 snapshot with 79.31% caselaw v2 score. BenchLM shows this benchmark for display only and does not use it in overall rankings.
How many models are evaluated on CaseLaw v2?
The May 4, 2026 snapshot contains 54 AI models.
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.