CaseLaw (Vals AI) leaderboard
- Grok 4.3 leads
- 79.3%
- models measured
- 45
- lead over #2
- +5.9 pts
- median
- 59.4%
Leader on CaseLaw (Vals AI) · read October 21
On October 21, 2026, 45 of our catalog's models had a CaseLaw (Vals AI) result, and SpaceXAI Grok 4.3 led with 79.3%.
Questions on Canadian case law from a private dataset. Share answered correctly, as run by Vals AI. An archived board.
Role in the index
We collect this benchmark and show it as supporting evidence. It is not part of the Vibe Coding Index, because it measures something other than coding and the index is about coding. Read the methodology
| # | Model | Result | Latency, as printed | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|---|
| 1 | Grok 4.3xAI | 79.3% | 43 s | 22.5 | $2.50 | |
| 2 | GPT-5.1OpenAI | 73.4% | 24 s | 53.3 | $10.00 | |
| 3 | GPT-4.1OpenAI | 69.9% | 6.7 s | 46.5 | $8.00 | |
| 4 | GPT-5 MiniOpenAI | 68.5% | 43 s | 59.8 | $2.00 | |
| 5 | Claude Opus 4.7Anthropic | 68.4% | 18 s | 51.1 | $25.00 | |
| 6 | GPT-5OpenAI | 66.5% | 2m 03s | 72.2 | $10.00 | |
| 7 | GPT-5.5OpenAI | 66.2% | 25 s | 67.8 | $30.00 | |
| 8 | GPT-5.2OpenAI | 66.0% | 1m 26s | 58.9 | $14.00 | |
| 9 | Kimi K2 ThinkingMoonshot AI | 65.7% | 1m 31s | 63.4 | $2.50 | |
| 10 | Gemini 3.1 Pro PreviewGoogle | 64.8% | 57 s | 35.3 | $12.00 | |
| 11 | Command ACohere | 64.5% | 42 s | Not benchmarked | $10.00 | |
| 12 | Claude Sonnet 4.6Anthropic | 64.0% | 36 s | 57.8 | $15.00 | |
| 13 | Gemini 2.5 ProGoogle | 63.9% | 44 s | 37.0 | $10.00 | |
| 14 | GPT-5.4OpenAI | 63.8% | 57 s | 37.7 | $15.00 | |
| 15 | Claude Opus 4.5Anthropic | 62.6% | 17 s | 66.8 | $25.00 | |
| 16 | Claude Sonnet 4.5Anthropic | 62.2% | 25 s | 59.1 | $15.00 | |
| 17 | Claude Opus 4.6Anthropic | 62.1% | 32 s | 66.9 | $25.00 | |
| 18 | Mistral Large 3 2512Mistral AI | 61.4% | 41 s | Not benchmarked | $1.50 | |
| 19 | Kimi K2.6Moonshot AI | 61.2% | 46 s | 44.2 | $3.41 | |
| 20 | MiniMax M2.7MiniMax | 60.9% | 22 s | 28.4 | $1.20 | |
| 21 | GPT-4o (2024-11-20)OpenAI | 59.7% | 41 s | 18.2 | $10.00 | |
| 22 | Qwen3.5 Plus 2026-04-20Qwen | 59.7% | 1m 35s | Not benchmarked | $1.80 | |
| 23 | DeepSeek V4 Pro 0423DeepSeek | 59.4% | 48 s | 35.3 | $1.9105 | |
| 24 | Kimi K2.5Moonshot AI | 58.7% | 4m 01s | 58.5 | $2.50 | |
| 25 | Trinity Large ThinkingArcee AI | 57.9% | 15 s | 5.4 | $0.80 | |
| 26 | Claude Haiku 4.5Anthropic | 56.5% | 33 s | 49.8 | $5.00 | |
| 27 | Qwen3.5-FlashQwen | 56.0% | 44 s | 6.2 | $0.26 | |
| 28 | Gemini 3 Flash PreviewGoogle | 55.8% | 4m 21s | 75.8 | $3.00 | |
| 29 | MiniMax M2.1MiniMax | 55.8% | 37 s | Not benchmarked | $1.20 | |
| 30 | DeepSeek V3.2DeepSeek | 55.4% | 1m 06s | 54.4 | $0.42 | |
| 31 | Gemini 3.1 Flash Lite PreviewGoogle | 55.0% | 4.6 s | 8.1 | $1.50 | |
| 32 | GLM 4.7Z.ai | 54.9% | 60 s | 33.6 | $1.98 | |
| 33 | Grok 4.20xAI | 54.4% | 11 s | Not benchmarked | $2.50 | |
| 34 | MiniMax M2.5MiniMax | 53.5% | 31 s | 59.4 | $0.95 | |
| 35 | Qwen3.6 27BQwen | 53.2% | 42 s | Not benchmarked | $2.00 | |
| 36 | Gemma 4 31BGoogle | 52.6% | 29 s | 38.3 | $0.40 | |
| 37 | GPT-5 NanoOpenAI | 52.6% | 56 s | 34.8 | $0.40 | |
| 38 | GLM 5Z.ai | 52.5% | 1m 58s | 59.7 | $3.20 | |
| 39 | GPT-5.4 NanoOpenAI | 51.9% | 7.6 s | Not benchmarked | $1.25 | |
| 40 | GPT-5.4 MiniOpenAI | 51.7% | 58 s | 28.2 | $4.50 | |
| 41 | GLM 5.1Z.ai | 51.5% | 49 s | 44.0 | $4.40 | |
| 42 | Qwen3.6 PlusQwen | 51.5% | 44 s | 37.4 | $1.95 | |
| 43 | gpt-oss-120bOpenAI | 48.8% | 1m 08s | 34.5 | $0.60 | |
| 44 | Mistral Medium 3.5Mistral AI | 44.2% | 1m 05s | 9.2 | $7.50 | |
| 45 | gpt-oss-20bOpenAI | 43.8% | 1m 26s | Not benchmarked | $0.09 |
Only models in our catalog are listed, and each appears once with its best published run. 132 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out. Every other column is a figure the board publishes for the same run.