Encyclo-K - Law: leaderboard

Encyclo-K's multi-statement selection questions restricted to Law: accuracy over the 189 Law questions of the 5,038-question standard set.

Metric: Accuracy on the 189 Law questions (%). Source: encyclo-k.github.io. Saturation forecast: Around May 2027. 37 models tracked. Updated 17 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1DeepSeek V3.1 TerminusThinking52.38#440
2Qwen 3 235B A22B 2507 Instruct47.62#618
3Qwen 3 235B A22B 2507 (Thinking)43.39#469
4Qwen 3 235B A22BThinking40.21#631
5Kimi K2 090536.51#564
6Qwen 3 30B A3B 2507 (Thinking)36.51#765
7Qwen 3 30B A3B 2507 Instruct35.45#1091
8QwQ-32B35.45#908
9DeepSeek V3 (0324)32.8#722
10Qwen 3 32BThinking31.22#965
11Qwen 3 14BReasoning28.04#1264
12Qwen 2.5 32B Instruct27.51#1257
13Qwen 3 30B A3BThinking27.51#1155
14Qwen 3 4B 2507 Instruct25.93#2215
15Qwen 2.5 72B Instruct25.4#1119

No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.

Interactive version: theaggregate.ai/benchmark?slug=encyclo-k-law · How It Works · Data refreshed daily, snapshot 2026-11-17.