Skip to content
Weights Not Walls
Menu

Mystery Game Puzzles

Re-run by Epoch AI

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

Kind
Run independently by Epoch AI, with public logs. Our own re-runs are not built yet.
Score shown
Epoch's column "Best score (across scorers)", as a percentage, higher is better; random guessing scores about 9.2%. Error bars as the source gives them.
Results
148 in all: 29 for open-weight models, 112 for closed ones, 7 where Epoch does not say. 26 linked to a model page here.

Best open-weight result: deepseek-ai/DeepSeek-V4.1-Flash-1017, 49.0%, number 19 of 148 results.

Mystery Game Puzzles results
#ModelWeightsSettingScoreDateSource
1gpt-6.2-astra_maxOpenAIclosedeffort max95.0%±2.227 Oct 2026Epoch AI
2claude-fable-5-2_maxAnthropicclosedeffort max90.0%±3.06 Nov 2026Epoch AI
3gpt-6-astra_maxOpenAIclosedeffort max84.0%±3.730 Aug 2026Epoch AI
4gpt-6.1-sol_maxOpenAIclosedeffort max80.0%±4.029 Sep 2026Epoch AI
5gpt-6.1-terra_maxOpenAIclosedeffort max74.0%±4.426 Oct 2026Epoch AI
6gemini-4-carbon_highGoogle DeepMindclosedeffort high73.0%±4.52 Nov 2026Epoch AI
7claude-opus-5-5_maxAnthropicclosedeffort max71.0%±4.629 Sep 2026Epoch AI
8claude-sonnet-5-5_maxAnthropicclosedeffort max65.0%±4.829 Sep 2026Epoch AI
9gemini-4-hydrogen_xhighGoogle DeepMindclosedeffort xhigh65.0%±4.813 Nov 2026Epoch AI
10gemini-4-argon_highGoogle DeepMindclosedeffort high64.0%±4.813 Oct 2026Epoch AI
11claude-opus-5_maxAnthropicclosedeffort max59.0%±4.925 Jul 2026Epoch AI
12claude-fable-5-1_maxAnthropicclosedeffort max58.0%±5.01 Sep 2026Epoch AI
13gpt-5.6-sol_maxOpenAIclosedeffort max58.0%±5.028 Jul 2026Epoch AI
14gpt-5.5_xhighOpenAIclosedeffort xhigh56.0%±5.024 Jul 2026Epoch AI
15gpt-6-sol_maxOpenAIclosedeffort max56.0%±5.022 Sep 2026Epoch AI
16claude-fable-5_maxAnthropicclosedeffort max52.0%±5.031 Jul 2026Epoch AI
17gpt-5.5_highOpenAIclosedeffort high52.0%±5.027 Jul 2026Epoch AI
18muse-wonder-1.0_maxMeta AIclosedeffort max51.0%±5.014 Oct 2026Epoch AI
19deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_max, DeepSeekopeneffort max49.0%±5.019 Oct 2026Epoch AI
20gemini-3.8-flash_highGoogle DeepMindclosedeffort high47.0%±5.02 Sep 2026Epoch AI
21deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_max, DeepSeekopeneffort max43.0%±5.019 Aug 2026Epoch AI
22deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_max, DeepSeekopeneffort max43.0%±5.08 Oct 2026Epoch AI
23moonshotai/Kimi-K3.1-ProEpoch: kimi-k3.1-pro-together_max, Moonshotopeneffort max42.0%±5.06 Nov 2026Epoch AI
24qwen3.8-max_xhighAlibabaclosedeffort xhigh38.0%±4.95 Aug 2026Epoch AI
25claude-opus-5Anthropiccloseddefault37.0%±4.96 Aug 2026Epoch AI log
26gemini-3.7-flash_highGoogle DeepMindclosedeffort high37.0%±4.914 Aug 2026Epoch AI
27gpt-5.4-2026-03-05_xhighOpenAIclosedeffort xhigh37.0%±4.924 Jul 2026Epoch AI
28claude-opus-4-8_maxAnthropicclosedeffort max36.0%±4.825 Jul 2026Epoch AI
29claude-sonnet-5_maxAnthropicclosedeffort max35.0%±4.828 Jul 2026Epoch AI
30gpt-5.6-terra_maxOpenAIclosedeffort max35.0%±4.828 Jul 2026Epoch AI
31deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_max, DeepSeekopeneffort max34.0%±4.85 Aug 2026Epoch AI
32gemini-3.1-pro-preview_highGoogle DeepMindclosedeffort high34.0%±4.827 Jul 2026Epoch AI
33grok-4.6_xhighxAIclosedeffort xhigh34.0%±4.814 Aug 2026Epoch AI
34zai-org/GLM-5.3Epoch: glm-5.3_max, Z.ai (Zhipu AI)openeffort max33.0%±4.730 Aug 2026Epoch AI
35gpt-5.6-sol_noneOpenAIclosedeffort none33.0%±4.727 Aug 2026Epoch AI log
36gemini-3.1-pro-preview_mediumGoogle DeepMindclosedeffort medium32.0%±4.76 Aug 2026Epoch AI log
37gemini-3.5-flash_highGoogle DeepMindclosedeffort high32.0%±4.727 Jul 2026Epoch AI
38qwen3.7-maxAlibabacloseddefault32.0%±4.728 Jul 2026Epoch AI
39claude-opus-4-8_xhighAnthropicclosedeffort xhigh31.0%±4.626 Jul 2026Epoch AI
40claude-haiku-5-5_maxAnthropicclosedeffort max30.0%±4.67 Oct 2026Epoch AI
41gemini-3.6-flash_highGoogle DeepMindclosedeffort high30.0%±4.65 Aug 2026Epoch AI
42gemini-3.1-pro-preview_lowGoogle DeepMindclosedeffort low29.0%±4.66 Aug 2026Epoch AI log
43grok-4.7_xhighxAIclosedeffort xhigh29.0%±4.622 Sep 2026Epoch AI
44o3-2025-04-16_highOpenAIclosedeffort high29.0%±4.627 Aug 2026Epoch AI log
45claude-opus-4-7_maxAnthropicclosedeffort max28.0%±4.525 Jul 2026Epoch AI
46gemini-3.5-flash_lowGoogle DeepMindclosedeffort low28.0%±4.55 Aug 2026Epoch AI log
47gpt-5.4-2026-03-05_mediumOpenAIclosedeffort medium28.0%±4.528 Aug 2026Epoch AI log
48gpt-5.5_lowOpenAIclosedeffort low28.0%±4.528 Aug 2026Epoch AI log
49gemini-3-flash-preview_lowGoogle DeepMindclosedeffort low26.0%±4.45 Aug 2026Epoch AI log
50gpt-5.6-sol_lowOpenAIclosedeffort low26.0%±4.427 Aug 2026Epoch AI log
51moonshotai/Kimi-K3Epoch: kimi-k3_max, Moonshotopeneffort max26.0%±4.429 Jul 2026Epoch AI
52qwen3.7-max_noneAlibabaclosedeffort none26.0%±4.427 Aug 2026Epoch AI log
53claude-opus-4-6_maxAnthropicclosedeffort max25.0%±4.425 Jul 2026Epoch AI
54gemini-3-flash-preview_minimalGoogle DeepMindclosedeffort minimal25.0%±4.45 Aug 2026Epoch AI log
55gemini-3.6-flash_minimalGoogle DeepMindclosedeffort minimal25.0%±4.45 Aug 2026Epoch AI log
56muse-spark-1.3_maxMeta AIclosedeffort max25.0%±4.518 Sep 2026Epoch AI
57gemini-3.6-flash_lowGoogle DeepMindclosedeffort low23.0%±4.25 Aug 2026Epoch AI log
58gpt-5-2025-08-07_highOpenAIclosedeffort high23.0%±4.25 Aug 2026Epoch AI
59gpt-5.2-2025-12-11_highOpenAIclosedeffort high23.0%±4.26 Aug 2026Epoch AI
60o3-2025-04-16_mediumOpenAIclosedeffort medium23.0%±4.227 Aug 2026Epoch AI log
61claude-opus-4-5-20251101_48KAnthropiccloseddefault22.0%±4.225 Jul 2026Epoch AI
62gpt-5.2-2025-12-11_mediumOpenAIclosedeffort medium22.0%±4.228 Aug 2026Epoch AI log
63Qwen/Qwen3.6-35B-A3BEpoch: qwen3.6-35b-a3b_none, Alibabaopeneffort none22.0%±4.227 Aug 2026Epoch AI log
64claude-opus-4-1-20250805_24KAnthropiccloseddefault21.0%±4.125 Jul 2026Epoch AI
65gpt-5.6-luna_maxOpenAIclosedeffort max21.0%±4.128 Jul 2026Epoch AI
66gemini-3-flash-preview_highGoogle DeepMindclosedeffort high20.0%±4.05 Aug 2026Epoch AI log
67gpt-5.6-luna_noneOpenAIclosedeffort none20.0%±4.030 Aug 2026Epoch AI log
68mistral-large-4-0-oss_highMistral AIno page matchopeneffort high20.0%±4.02 Nov 2026Epoch AI
69nemotron-3-ultraNvidiano page matchopendefault20.0%±4.027 Aug 2026Epoch AI log
70qwen3.5-flashAlibabanot stateddefault20.0%±4.027 Aug 2026Epoch AI log
71gemini-3.5-flash-lite_lowGoogle DeepMindclosedeffort low19.0%±3.95 Aug 2026Epoch AI log
72zai-org/GLM-5.2Epoch: glm-5.2_low, Z.ai (Zhipu AI)openeffort low19.0%±3.927 Aug 2026Epoch AI log
73gpt-5.1-2025-11-13_lowOpenAIclosedeffort low19.0%±3.927 Aug 2026Epoch AI log
74gpt-5.6-terra_lowOpenAIclosedeffort low19.0%±3.927 Aug 2026Epoch AI log
75qwen3.6-max-preview_noneAlibabanot statedeffort none19.0%±3.927 Aug 2026Epoch AI log
76zai-org/GLM-5.2Epoch: glm-5.2_minimal, Z.ai (Zhipu AI)openeffort minimal18.0%±3.927 Aug 2026Epoch AI log
77zai-org/GLM-5.2Epoch: glm-5.2_none, Z.ai (Zhipu AI)openeffort none18.0%±3.927 Aug 2026Epoch AI log
78gpt-5.5_noneOpenAIclosedeffort none18.0%±3.927 Aug 2026Epoch AI log
79moonshotai/Kimi-K2.6Epoch: kimi-k2.6, Moonshotopendefault18.0%±3.917 Jul 2026Epoch AI
80Qwen/Qwen3.5-397B-A17BEpoch: qwen3.5-397b-a17b_none, Alibabaopeneffort none18.0%±3.927 Aug 2026Epoch AI log
81qwen3.6-flash_noneAlibabanot statedeffort none18.0%±3.927 Aug 2026Epoch AI log
82claude-sonnet-4-5-20250929_48KAnthropiccloseddefault17.0%±3.825 Jul 2026Epoch AI
83deepseek-ai/DeepSeek-V4-ProEpoch: deepseek-v4-pro_none, DeepSeekopeneffort none17.0%±3.827 Aug 2026Epoch AI log
84gpt-5.4-2026-03-05_lowOpenAIclosedeffort low17.0%±3.827 Aug 2026Epoch AI log
85gpt-5.6-luna_lowOpenAIclosedeffort low17.0%±3.827 Aug 2026Epoch AI log
86mistral-large-4-0_highMistral AIno page matchopeneffort high17.0%±3.813 Oct 2026Epoch AI
87Qwen/Qwen3.5-122B-A10BEpoch: qwen3.5-122b-a10b_none, Alibabaopeneffort none17.0%±3.827 Aug 2026Epoch AI log
88qwen3.5-plus_noneAlibabanot statedeffort none17.0%±3.827 Aug 2026Epoch AI log
89qwen3.7-plusAlibabacloseddefault17.0%±3.828 Aug 2026Epoch AI log
90qwen3.7-plus_noneAlibabaclosedeffort none17.0%±3.827 Aug 2026Epoch AI log
91claude-sonnet-4-6_lowAnthropicclosedeffort low16.0%±3.76 Aug 2026Epoch AI log
92claude-sonnet-5Anthropiccloseddefault16.0%±3.76 Aug 2026Epoch AI log
93gpt-5-2025-08-07_lowOpenAIclosedeffort low16.0%±3.727 Aug 2026Epoch AI log
94gpt-5.1-2025-11-13_mediumOpenAIclosedeffort medium16.0%±3.727 Aug 2026Epoch AI log
95gpt-5.4-2026-03-05_noneOpenAIclosedeffort none16.0%±3.730 Aug 2026Epoch AI log
96qwen3.5-flash_noneAlibabanot statedeffort none16.0%±3.727 Aug 2026Epoch AI log
97qwen3.5-plusAlibabanot stateddefault16.0%±3.75 Aug 2026Epoch AI log
98claude-opus-4-6Anthropiccloseddefault15.0%±3.66 Aug 2026Epoch AI log
99zai-org/GLM-5.2Epoch: glm-5.2_medium, Z.ai (Zhipu AI)openeffort medium15.0%±3.627 Aug 2026Epoch AI log
100gpt-5-2025-08-07_mediumOpenAIclosedeffort medium15.0%±3.627 Aug 2026Epoch AI log
101gpt-5.1-2025-11-13_noneOpenAIclosedeffort none15.0%±3.627 Aug 2026Epoch AI log
102qwen3.7-flashAlibabacloseddefault15.0%±3.628 Aug 2026Epoch AI log
103muse-spark-1.3_xhighMeta AIclosedeffort xhigh14.1%±3.516 Sep 2026Epoch AI
104claude-sonnet-4-6Anthropiccloseddefault14.0%±3.56 Aug 2026Epoch AI log
105gpt-5-2025-08-07_minimalOpenAIclosedeffort minimal14.0%±3.527 Aug 2026Epoch AI log
106gpt-5.2-2025-12-11_noneOpenAIclosedeffort none14.0%±3.527 Aug 2026Epoch AI log
107gpt-5.6-terra_noneOpenAIclosedeffort none14.0%±3.527 Aug 2026Epoch AI log
108qwen3.7-flash_noneAlibabaclosedeffort none14.0%±3.527 Aug 2026Epoch AI log
109claude-opus-4-7Anthropiccloseddefault13.0%±3.46 Aug 2026Epoch AI log
110gemini-3.5-flash-lite_minimalGoogle DeepMindclosedeffort minimal12.0%±3.35 Aug 2026Epoch AI log
111gpt-4-0613OpenAIcloseddefault12.0%±3.328 Aug 2026Epoch AI log
112gpt-4o-mini-2024-07-18OpenAIcloseddefault12.0%±3.327 Aug 2026Epoch AI log
113gpt-5.6-luna_mediumOpenAIclosedeffort medium12.0%±3.327 Aug 2026Epoch AI log
114moonshotai/Kimi-K2.6Epoch: kimi-k2.6_none, Moonshotopeneffort none12.0%±3.329 Aug 2026Epoch AI log
115moonshotai/Kimi-K3Epoch: kimi-k3_low, Moonshotopeneffort low12.0%±3.329 Aug 2026Epoch AI log
116qwen3.6-flashAlibabanot stateddefault12.0%±3.327 Aug 2026Epoch AI log
117qwen3.6-plus_noneAlibabaclosedeffort none12.0%±3.327 Aug 2026Epoch AI log
118gpt-5.4-mini-2026-03-17_noneOpenAIclosedeffort none11.0%±3.127 Aug 2026Epoch AI log
119gpt-5-mini-2025-08-07_minimalOpenAIclosedeffort minimal10.0%±3.027 Aug 2026Epoch AI log
120gpt-5.2-2025-12-11_lowOpenAIclosedeffort low10.0%±3.027 Aug 2026Epoch AI log
121gpt-5.6-terra_mediumOpenAIclosedeffort medium10.0%±3.028 Aug 2026Epoch AI log
122gpt-5-nano-2025-08-07_lowOpenAIclosedeffort low9.0%±2.927 Aug 2026Epoch AI log
123gpt-5-nano-2025-08-07_mediumOpenAIclosedeffort medium9.0%±2.927 Aug 2026Epoch AI log
124gpt-5.4-nano-2026-03-17_noneOpenAIclosedeffort none9.0%±2.927 Aug 2026Epoch AI log
125Qwen/Qwen3-235B-A22B-Thinking-2507Epoch: Qwen3-235B-A22B-Thinking-2507, Alibabaopendefault9.0%±2.927 Aug 2026Epoch AI log
126zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_max, Z.ai (Zhipu AI)openeffort max8.0%±2.728 Aug 2026Epoch AI
127gpt-5-nano-2025-08-07_highOpenAIclosedeffort high8.0%±2.727 Aug 2026Epoch AI log
128gpt-5.4-mini-2026-03-17_lowOpenAIclosedeffort low8.0%±2.727 Aug 2026Epoch AI log
129MiniMaxAI/MiniMax-M3Epoch: MiniMax-M3_none, MiniMaxopeneffort none8.0%±2.727 Aug 2026Epoch AI log
130claude-opus-4-6_lowAnthropicclosedeffort low7.0%±2.66 Aug 2026Epoch AI log
131gpt-4.1-mini-2025-04-14OpenAIcloseddefault7.0%±2.627 Aug 2026Epoch AI log
132gpt-5.4-mini-2026-03-17_mediumOpenAIclosedeffort medium7.0%±2.627 Aug 2026Epoch AI log
133gpt-6-luna_maxOpenAIclosedeffort max7.0%±2.622 Sep 2026Epoch AI
134MiniMaxAI/MiniMax-M3Epoch: MiniMax-M3, MiniMaxopendefault7.0%±2.627 Aug 2026Epoch AI log
135o3-mini-2025-01-31_highOpenAIclosedeffort high7.0%±2.627 Aug 2026Epoch AI log
136Qwen/Qwen3.6-27BEpoch: qwen3.6-27b_none, Alibabaopeneffort none7.0%±2.627 Aug 2026Epoch AI log
137gpt-5.4-nano-2026-03-17_mediumOpenAIclosedeffort medium6.0%±2.427 Aug 2026Epoch AI log
138thinkingmachines/Inkling-SmallEpoch: inkling-small_xhigh, Thinking Machinesopeneffort xhigh6.0%±2.415 Aug 2026Epoch AI
139gpt-5-mini-2025-08-07_highOpenAIclosedeffort high5.0%±2.227 Aug 2026Epoch AI log
140gpt-5-nano-2025-08-07_minimalOpenAIclosedeffort minimal5.0%±2.227 Aug 2026Epoch AI log
141gpt-5.4-nano-2026-03-17_highOpenAIclosedeffort high5.0%±2.227 Aug 2026Epoch AI log
142o4-mini-2025-04-16_highOpenAIclosedeffort high5.0%±2.227 Aug 2026Epoch AI log
143qwen3-max-2025-09-23Alibabacloseddefault5.0%±2.227 Aug 2026Epoch AI log
144gpt-5-mini-2025-08-07_mediumOpenAIclosedeffort medium4.0%±2.027 Aug 2026Epoch AI log
145gpt-3.5-turbo-0125OpenAIcloseddefault3.0%±1.727 Aug 2026Epoch AI log
146gpt-5.4-nano-2026-03-17_lowOpenAIclosedeffort low3.0%±1.727 Aug 2026Epoch AI log
147openai/gpt-oss-120bEpoch: gpt-oss-120b_medium, OpenAIopeneffort medium2.0%±1.427 Aug 2026Epoch AI log
148openai/gpt-oss-120bEpoch: gpt-oss-120b_high, OpenAIopeneffort high0.0%27 Aug 2026Epoch AI log

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