Skip to content
Weights Not Walls
Menu

CL-bench

Reported

No description yet: we only describe an eval set when we can do it correctly.

Kind
Published by a third party's leaderboard, collected by Epoch AI. The leaderboard is named and linked.
Score shown
Epoch's column "Overall", as a percentage, higher is better; random guessing scores about 0.0%. The source gives no error bars.
Results
23 in all: 8 for open-weight models, 14 for closed ones, 1 where Epoch does not say. 8 linked to a model page here.

Best open-weight result: moonshotai/Kimi-K2.5, 19.3%, number 10 of 23 results.

CL-bench results
#ModelWeightsSettingScoreDateSource
1gpt-5.4-2026-03-05_xhighOpenAIclosedeffort xhigh27.9%not datedvia Epoch AI
2gpt-5.1-2025-11-13_highOpenAIclosedeffort high23.7%not datedvia Epoch AI
3grok-4-20xAIcloseddefault22.2%not datedvia Epoch AI
4claude-opus-4-5-20251101_unknownAnthropicclosedeffort unknown21.1%not datedvia Epoch AI
5gpt-5.1-2025-11-13_unknownOpenAIclosedeffort unknown21.1%not datedvia Epoch AI
6gemini-3.1-pro-previewGoogle DeepMindcloseddefault20.8%not datedvia Epoch AI
7claude-opus-4-6_unknownAnthropicclosedeffort unknown20.7%not datedvia Epoch AI
8qwen3.6-plusAlibabacloseddefault20.3%not datedvia Epoch AI
9qwen3.5-plusAlibabanot stateddefault19.8%not datedvia Epoch AI
10moonshotai/Kimi-K2.5Epoch: kimi-k2.5, Moonshotopendefault19.3%not datedvia Epoch AI
11zai-org/GLM-5Epoch: glm-5, Z.ai (Zhipu AI)opendefault18.7%not datedvia Epoch AI
12gpt-5.2-2025-12-11_unknownOpenAIclosedeffort unknown18.2%not datedvia Epoch AI
13gpt-5.2-2025-12-11_highOpenAIclosedeffort high18.1%not datedvia Epoch AI
14o3-2025-04-16_highOpenAIclosedeffort high17.8%not datedvia Epoch AI
15moonshotai/Kimi-K2-ThinkingEpoch: moonshotai/Kimi-K2-Thinking, Moonshotopendefault17.6%not datedvia Epoch AI
16zai-org/GLM-4.7Epoch: zai-org/glm-4.7, Z.ai (Zhipu AI)opendefault15.9%not datedvia Epoch AI
17gemini-3-pro-previewGoogle DeepMindcloseddefault15.8%not datedvia Epoch AI
18mimo-v2-proXiaomi Corpcloseddefault15.7%not datedvia Epoch AI
19qwen3-max-2025-09-23Alibabacloseddefault14.5%not datedvia Epoch AI
20deepseek-ai/DeepSeek-V3.2-ExpEpoch: DeepSeek-V3.2-Exp_thinking, DeepSeekopeneffort thinking13.2%not datedvia Epoch AI
21deepseek-ai/DeepSeek-V3.2Epoch: deepseek/deepseek-v3.2, DeepSeekopendefault12.4%not datedvia Epoch AI
22moonshotai/Kimi-K2-ThinkingEpoch: kimi-k2-thinking, Moonshotopendefault11.9%not datedvia Epoch AI
23MiniMaxAI/MiniMax-M2.5Epoch: MiniMax-M2.5, MiniMaxopendefault11.4%not datedvia Epoch AI

Open-weight rows are shaded. Dates are when the result was run or evaluated, where the data says. Data from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Epoch AI. Method