Skip to content
Weights Not Walls
Menu

ARC-AGI

Reported

Visual puzzles on small coloured grids: from a few examples the model must find the rule and apply it to a new grid.

Kind
Published by a third party's leaderboard, collected by Epoch AI. The leaderboard is named and linked.
Link
https://arcprize.org/leaderboard
Score shown
Epoch's column "Score", as a percentage, higher is better; random guessing scores about 0.0%. The source gives no error bars.
Results
318 in all: 53 for open-weight models, 263 for closed ones, 2 where Epoch does not say. 49 linked to a model page here.

Best open-weight result: thinkingmachines/Inkling-2-Small, 96.5%, number 39 of 318 results.

ARC-AGI results
#ModelWeightsSettingScoreDateSource
1claude-fable-5_maxAnthropicclosedeffort max98.5%not datedarcprize.org
2claude-fable-5_xhighAnthropicclosedeffort xhigh98.5%not datedarcprize.org
3claude-fable-5-2_xhighAnthropicclosedeffort xhigh98.5%not datedarcprize.org
4claude-opus-5-5_highAnthropicclosedeffort high98.5%not datedarcprize.org
5gemini-3.8-flash_highGoogle DeepMindclosedeffort high98.5%not datedarcprize.org
6gemini-4-argon_highGoogle DeepMindclosedeffort high98.5%not datedarcprize.org
7gemini-4-argon_mediumGoogle DeepMindclosedeffort medium98.5%not datedarcprize.org
8gemini-4-carbon_mediumGoogle DeepMindclosedeffort medium98.5%not datedarcprize.org
9gpt-6-astra_highOpenAIclosedeffort high98.5%not datedarcprize.org
10gpt-6-astra_xhighOpenAIclosedeffort xhigh98.5%not datedarcprize.org
11gpt-6.1-sol_highOpenAIclosedeffort high98.5%not datedarcprize.org
12gpt-6.1-sol_xhighOpenAIclosedeffort xhigh98.5%not datedarcprize.org
13gemini-3.1-pro-previewGoogle DeepMindcloseddefault98.0%not datedARC Prize Leaderboard
14claude-fable-5-1_maxAnthropicclosedeffort max97.5%not datedarcprize.org
15claude-fable-5-2_highAnthropicclosedeffort high97.5%not datedarcprize.org
16claude-fable-5-2_maxAnthropicclosedeffort max97.5%not datedarcprize.org
17claude-fable-5-2_mediumAnthropicclosedeffort medium97.5%not datedarcprize.org
18claude-opus-5_highAnthropicclosedeffort high97.5%not datedarcprize.org
19claude-opus-5_maxAnthropicclosedeffort max97.5%not datedarcprize.org
20claude-opus-5_maxAnthropicclosedeffort max97.5%not datedarcprize.org
21claude-opus-5-5_maxAnthropicclosedeffort max97.5%not datedarcprize.org
22claude-opus-5-5_mediumAnthropicclosedeffort medium97.5%not datedarcprize.org
23claude-opus-5-5_xhighAnthropicclosedeffort xhigh97.5%not datedarcprize.org
24gemini-3.8-flash_mediumGoogle DeepMindclosedeffort medium97.5%not datedarcprize.org
25gemini-4-carbon_highGoogle DeepMindclosedeffort high97.5%not datedarcprize.org
26gpt-5.6-sol_xhighOpenAIclosedeffort xhigh97.5%not datedarcprize.org
27gpt-5.6-sol_xhighOpenAIclosedeffort xhigh97.5%not datedarcprize.org
28gpt-6-astra_maxOpenAIclosedeffort max97.5%not datedarcprize.org
29gpt-6-astra_mediumOpenAIclosedeffort medium97.5%not datedarcprize.org
30gpt-5.6-sol_highOpenAIclosedeffort high97.0%not datedarcprize.org
31claude-fable-5-1_xhighAnthropicclosedeffort xhigh96.5%not datedarcprize.org
32claude-sonnet-5-5_maxAnthropicclosedeffort max96.5%not datedarcprize.org
33claude-sonnet-5-5_xhighAnthropicclosedeffort xhigh96.5%not datedarcprize.org
34gpt-5.5-pro_highOpenAIclosedeffort high96.5%not datedarcprize.org
35gpt-5.6-sol_maxOpenAIclosedeffort max96.5%not datedarcprize.org
36gpt-5.6-terra_maxOpenAIclosedeffort max96.5%not datedarcprize.org
37gpt-6-astra_lowOpenAIclosedeffort low96.5%not datedarcprize.org
38gpt-6.1-sol_maxOpenAIclosedeffort max96.5%not datedarcprize.org
39thinkingmachines/Inkling-2-SmallEpoch: inkling-2-small_xhigh, Thinking Machinesopeneffort xhigh96.5%not datedarcprize.org
40claude-fable-5-1_highAnthropicclosedeffort high96.0%not datedarcprize.org
41gemini-3-deep-think-previewGoogle DeepMindcloseddefault96.0%not datedarcprize.org
42claude-fable-5_highAnthropicclosedeffort high95.5%not datedarcprize.org
43gemini-3.7-flash_highGoogle DeepMindclosedeffort high95.5%not datedarcprize.org
44gpt-6-sol_maxOpenAIclosedeffort max95.5%not datedarcprize.org
45gpt-6.1-sol_mediumOpenAIclosedeffort medium95.5%not datedarcprize.org
46gpt-5.5_xhighOpenAIclosedeffort xhigh95.0%not datedarcprize.org
47gpt-5.5_xhighOpenAIclosedeffort xhigh95.0%not datedarcprize.org
48gpt-5.5-pro_xhighOpenAIclosedeffort xhigh95.0%not datedarcprize.org
49gpt-5.5-pro_xhighOpenAIclosedeffort xhigh95.0%not datedarcprize.org
50claude-fable-5-1_mediumAnthropicclosedeffort medium94.5%not datedarcprize.org
51claude-fable-5-2_lowAnthropicclosedeffort low94.5%not datedarcprize.org
52deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_max, DeepSeekopeneffort max94.5%not datedarcprize.org
53deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_high, DeepSeekopeneffort high94.5%not datedarcprize.org
54gpt-5.4-pro-2026-03-05_xhighOpenAIclosedeffort xhigh94.5%not datedvia Epoch AI
55gpt-5.4-pro-2026-03-05_xhighOpenAIclosedeffort xhigh94.5%not datedarcprize.org
56gpt-5.5_highOpenAIclosedeffort high94.5%not datedarcprize.org
57moonshotai/Kimi-K3Epoch: kimi-k3_max, Moonshotopeneffort max94.5%not datedarcprize.org
58claude-opus-4-6_120KAnthropiccloseddefault94.0%not datedarcprize.org
59claude-opus-4-6_120KAnthropiccloseddefault94.0%not datedARC Prize Leaderboard
60gpt-5.6-terra_xhighOpenAIclosedeffort xhigh94.0%not datedarcprize.org
61gpt-5.6-terra_xhighOpenAIclosedeffort xhigh94.0%not datedarcprize.org
62gpt-5.4-2026-03-05_xhighOpenAIclosedeffort xhigh93.7%not datedvia Epoch AI
63gpt-5.4-2026-03-05_xhighOpenAIclosedeffort xhigh93.7%not datedarcprize.org
64claude-opus-4-7_highAnthropicclosedeffort high93.5%not datedvia Epoch AI
65claude-opus-4-7_highAnthropicclosedeffort high93.5%not datedarcprize.org
66deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_max, DeepSeekopeneffort max93.5%not datedarcprize.org
67gpt-6.1-sol_lowOpenAIclosedeffort low93.5%not datedarcprize.org
68claude-opus-4-6_120KAnthropiccloseddefault93.0%not datedarcprize.org
69gpt-5.4-2026-03-05_highOpenAIclosedeffort high92.7%not datedvia Epoch AI
70gpt-6-sol_xhighOpenAIclosedeffort xhigh92.7%not datedarcprize.org
71claude-fable-5_mediumAnthropicclosedeffort medium92.5%not datedarcprize.org
72claude-opus-4-8_maxAnthropicclosedeffort max92.5%not datedarcprize.org
73gemini-3.5-flash_highGoogle DeepMindclosedeffort high92.5%not datedarcprize.org
74gpt-5.6-sol_mediumOpenAIclosedeffort medium92.5%not datedarcprize.org
75gpt-5.5_mediumOpenAIclosedeffort medium92.2%not datedarcprize.org
76claude-opus-4-6_120KAnthropiccloseddefault92.0%not datedarcprize.org
77claude-opus-4-7_maxAnthropicclosedeffort max92.0%not datedvia Epoch AI
78claude-opus-4-7_maxAnthropicclosedeffort max92.0%not datedarcprize.org
79claude-opus-4-8_highAnthropicclosedeffort high92.0%not datedarcprize.org
80claude-sonnet-5-5_highAnthropicclosedeffort high92.0%not datedarcprize.org
81gpt-5.6-terra_highOpenAIclosedeffort high92.0%not datedarcprize.org
82claude-opus-4-8_mediumAnthropicclosedeffort medium91.5%not datedarcprize.org
83deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_low, DeepSeekopeneffort low91.5%not datedarcprize.org
84gemini-3.6-flash_highGoogle DeepMindclosedeffort high91.2%not datedarcprize.org
85gemini-3.7-flash_mediumGoogle DeepMindclosedeffort medium91.2%not datedarcprize.org
86claude-opus-4-7_lowAnthropicclosedeffort low91.0%not datedarcprize.org
87claude-opus-4-7_mediumAnthropicclosedeffort medium91.0%not datedarcprize.org
88zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_max, Z.ai (Zhipu AI)openeffort max91.0%not datedarcprize.org
89gpt-6-sol_highOpenAIclosedeffort high91.0%not datedarcprize.org
90claude-fable-5_lowAnthropicclosedeffort low90.5%not datedarcprize.org
91deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_low, DeepSeekopeneffort low90.5%not datedarcprize.org
92deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_low, DeepSeekopeneffort low90.5%not datedarcprize.org
93gemini-3.8-flash_lowGoogle DeepMindclosedeffort low90.5%not datedarcprize.org
94gpt-5.2-pro-2025-12-11_xhighOpenAIclosedeffort xhigh90.5%not datedvia Epoch AI
95gpt-5.2-pro-2025-12-11_xhighOpenAIclosedeffort xhigh90.5%not datedarcprize.org
96gpt-5.2-pro-2025-12-11_xhighOpenAIclosedeffort xhigh90.5%not datedarcprize.org
97grok-4.7_highxAIclosedeffort high90.2%not datedarcprize.org
98claude-fable-5-1_lowAnthropicclosedeffort low90.0%not datedarcprize.org
99deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_max, DeepSeekopeneffort max90.0%not datedarcprize.org
100grok-4.7_mediumxAIclosedeffort medium90.0%not datedarcprize.org
101grok-4-20xAIcloseddefault89.5%not datedarcprize.org
102grok-4.7_xhighxAIclosedeffort xhigh89.5%not datedarcprize.org
103claude-haiku-5-5_maxAnthropicclosedeffort max89.0%not datedarcprize.org
104deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_max, DeepSeekopeneffort max89.0%not datedarcprize.org
105meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_xhigh, Meta AIopeneffort xhigh89.0%not datedarcprize.org
106claude-opus-5-5_lowAnthropicclosedeffort low88.5%not datedarcprize.org
107deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_high, DeepSeekopeneffort high88.5%not datedarcprize.org
108meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_high, Meta AIopeneffort high88.5%not datedarcprize.org
109claude-opus-4-8_lowAnthropicclosedeffort low88.0%not datedarcprize.org
110gpt-5.6-luna_maxOpenAIclosedeffort max88.0%not datedarcprize.org
111gpt-5.6-luna_xhighOpenAIclosedeffort xhigh87.7%not datedarcprize.org
112gpt-5.6-luna_xhighOpenAIclosedeffort xhigh87.7%not datedarcprize.org
113claude-haiku-5-5_xhighAnthropicclosedeffort xhigh87.5%not datedarcprize.org
114gemini-3-deep-think-previewGoogle DeepMindcloseddefault87.5%not datedvia Epoch AI
115gemini-3-deep-think-previewGoogle DeepMindcloseddefault87.5%not datedarcprize.org
116grok-4.6_mediumxAIclosedeffort medium87.5%not datedarcprize.org
117Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_xhigh, Alibabaopeneffort xhigh87.5%not datedarcprize.org
118grok-4.5_mediumxAIclosedeffort medium87.2%not datedarcprize.org
119deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_high, DeepSeekopeneffort high87.2%not datedarcprize.org
120deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_high, DeepSeekopeneffort high87.0%not datedarcprize.org
121grok-4.6_highxAIclosedeffort high87.0%not datedarcprize.org
122grok-4.6_xhighxAIclosedeffort xhigh87.0%not datedarcprize.org
123gpt-6-luna_maxOpenAIclosedeffort max86.7%not datedarcprize.org
124moonshotai/Kimi-K3Epoch: kimi-k3_high, Moonshotopeneffort high86.7%not datedarcprize.org
125claude-sonnet-4-6_highAnthropicclosedeffort high86.5%not datedvia Epoch AI
126gpt-5.2-2025-12-11_xhighOpenAIclosedeffort xhigh86.2%not datedARC Prize Leaderboard
127gpt-5.2-2025-12-11_xhighOpenAIclosedeffort xhigh86.2%not datedarcprize.org
128gpt-5.2-2025-12-11_xhighOpenAIclosedeffort xhigh86.2%not datedarcprize.org
129gpt-5.4-2026-03-05_mediumOpenAIclosedeffort medium86.2%not datedvia Epoch AI
130claude-opus-4-6_120KAnthropiccloseddefault86.0%not datedarcprize.org
131claude-sonnet-4-6_maxAnthropicclosedeffort max86.0%not datedvia Epoch AI
132gpt-6-astra_noneOpenAIclosedeffort none86.0%not datedarcprize.org
133gpt-5.2-pro-2025-12-11_highOpenAIclosedeffort high85.7%not datedvia Epoch AI
134grok-4.5_highxAIclosedeffort high85.7%not datedarcprize.org
135gemini-4-carbon_lowGoogle DeepMindclosedeffort low85.5%not datedarcprize.org
136gemini-3.7-flash_lowGoogle DeepMindclosedeffort low85.2%not datedarcprize.org
137gemini-3-flash-previewGoogle DeepMindcloseddefault84.7%not datedARC Prize Leaderboard
138deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_low, DeepSeekopeneffort low84.0%not datedarcprize.org
139thinkingmachines/Inkling-SmallEpoch: inkling-small_xhigh, Thinking Machinesopeneffort xhigh84.0%not datedarcprize.org
140gpt-6-sol_mediumOpenAIclosedeffort medium83.7%not datedarcprize.org
141gemini-3.6-flash_mediumGoogle DeepMindclosedeffort medium83.2%not datedarcprize.org
142claude-haiku-5-5_highAnthropicclosedeffort high81.5%not datedarcprize.org
143gpt-5.2-pro-2025-12-11_mediumOpenAIclosedeffort medium81.2%not datedvia Epoch AI
144claude-opus-4-5-20251101_64KAnthropiccloseddefault80.0%not datedARC Prize Leaderboard
145claude-opus-4-5-20251101_64KAnthropiccloseddefault80.0%not datedarcprize.org
146thinkingmachines/InklingEpoch: Inkling, Thinking Machinesopendefault79.5%not datedarcprize.org
147grok-4.5_lowxAIclosedeffort low79.2%not datedarcprize.org
148gpt-5.2-2025-12-11_highOpenAIclosedeffort high78.7%not datedarcprize.org
149thinkingmachines/Inkling-SmallEpoch: inkling-small_high, Thinking Machinesopeneffort high78.0%not datedarcprize.org
150meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_medium, Meta AIopeneffort medium77.7%not datedarcprize.org
151zai-org/GLM-5.2Epoch: glm-5.2_unknown, Z.ai (Zhipu AI)openeffort unknown77.0%not datedarcprize.org
152gpt-5.6-terra_mediumOpenAIclosedeffort medium77.0%not datedarcprize.org
153gemini-3.6-flash_lowGoogle DeepMindclosedeffort low76.5%not datedarcprize.org
154gemini-4-argon_lowGoogle DeepMindclosedeffort low76.5%not datedarcprize.org
155gpt-5.6-luna_highOpenAIclosedeffort high76.5%not datedarcprize.org
156gpt-5.5_lowOpenAIclosedeffort low76.2%not datedarcprize.org
157claude-opus-4-5-20251101_32KAnthropiccloseddefault75.8%not datedarcprize.org
158claude-opus-4-5-20251101_32KAnthropiccloseddefault75.8%not datedARC Prize Leaderboard
159gemini-3-pro-previewGoogle DeepMindcloseddefault75.0%not datedARC Prize Leaderboard
160grok-4.6_lowxAIclosedeffort low74.8%not datedarcprize.org
161gpt-5.6-sol_lowOpenAIclosedeffort low74.5%not datedarcprize.org
162beam-1.0_xhighReflectionno page matchopeneffort xhigh74.0%not datedarcprize.org
163gpt-6-luna_xhighOpenAIclosedeffort xhigh73.0%not datedarcprize.org
164gpt-5.1-2025-11-13_highOpenAIclosedeffort high72.8%not datedarcprize.org
165gpt-5.1-2025-11-13_highOpenAIclosedeffort high72.8%not datedARC Prize Leaderboard
166gpt-5.2-2025-12-11_mediumOpenAIclosedeffort medium72.7%not datedvia Epoch AI
167gpt-6-sol_lowOpenAIclosedeffort low72.2%not datedarcprize.org
168claude-opus-4-5-20251101_16KAnthropiccloseddefault72.0%not datedvia Epoch AI
169zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_high, Z.ai (Zhipu AI)openeffort high71.8%not datedarcprize.org
170gpt-6-luna_highOpenAIclosedeffort high70.3%not datedarcprize.org
171gpt-5-pro-2025-10-06_highOpenAIclosedeffort high70.2%not datedARC Prize Leaderboard
172gpt-5-pro-2025-10-06_unknownOpenAIclosedeffort unknown70.2%not datedvia Epoch AI
173claude-sonnet-5-5_mediumAnthropicclosedeffort medium70.0%not datedarcprize.org
174Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_low, Alibabaopeneffort low69.2%not datedarcprize.org
175Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_medium, Alibabaopeneffort medium68.7%not datedarcprize.org
176gpt-5.4-2026-03-05_lowOpenAIclosedeffort low68.2%not datedvia Epoch AI
177beam-1.0_highReflectionno page matchopeneffort high67.7%not datedarcprize.org
178claude-sonnet-5-5_lowAnthropicclosedeffort low67.0%not datedarcprize.org
179thinkingmachines/Inkling-SmallEpoch: inkling-small_medium, Thinking Machinesopeneffort medium67.0%not datedarcprize.org
180grok-4-0709xAIcloseddefault66.7%not datedvia Epoch AI
181gpt-5-2025-08-07_highOpenAIclosedeffort high65.7%not datedarcprize.org
182moonshotai/Kimi-K3Epoch: kimi-k3_low, Moonshotopeneffort low65.7%not datedarcprize.org
183moonshotai/Kimi-K2.5Epoch: kimi-k2.5, Moonshotopendefault65.3%not datedvia Epoch AI
184claude-sonnet-4-5-20250929_32KAnthropiccloseddefault63.7%not datedvia Epoch AI
185gpt-5.4-mini-2026-03-17_xhighOpenAIclosedeffort xhigh63.7%not datedarcprize.org
186gpt-5.4-mini-2026-03-17_xhighOpenAIclosedeffort xhigh63.7%not datedarcprize.org
187MiniMaxAI/MiniMax-M2.5Epoch: MiniMax-M2.5, MiniMaxopendefault63.7%not datedvia Epoch AI
188grok-4.7_lowxAIclosedeffort low63.3%not datedarcprize.org
189gpt-6-luna_mediumOpenAIclosedeffort medium61.0%not datedarcprize.org
190o3-2025-04-16_highOpenAIclosedeffort high60.8%not datedarcprize.org
191gpt-5.6-terra_lowOpenAIclosedeffort low60.2%not datedarcprize.org
192o3-pro-2025-06-10_highOpenAIclosedeffort high59.3%not datedarcprize.org
193claude-opus-4-5-20251101_8KAnthropiccloseddefault58.7%not datedvia Epoch AI
194o4-mini-2025-04-16_highOpenAIclosedeffort high58.7%not datedarcprize.org
195meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_low, Meta AIopeneffort low58.5%not datedarcprize.org
196gpt-5.4-mini-2026-03-17_highOpenAIclosedeffort high58.0%not datedarcprize.org
197gpt-5.1-2025-11-13_mediumOpenAIclosedeffort medium57.7%not datedARC Prize Leaderboard
198gemini-3-flash-previewGoogle DeepMindcloseddefault57.7%not datedARC Prize Leaderboard
199gpt-5.1-2025-11-13_mediumOpenAIclosedeffort medium57.7%not datedarcprize.org
200claude-haiku-5-5_mediumAnthropicclosedeffort medium57.3%not datedarcprize.org
201deepseek-ai/DeepSeek-V3.2Epoch: deepseek/deepseek-v3.2, DeepSeekopendefault57.0%not datedvia Epoch AI
202o3-pro-2025-06-10_mediumOpenAIclosedeffort medium57.0%not datedARC Prize Leaderboard
203gpt-5.6-luna_mediumOpenAIclosedeffort medium56.5%not datedarcprize.org
204gpt-5-2025-08-07_mediumOpenAIclosedeffort medium56.2%not datedarcprize.org
205gpt-5.2-2025-12-11_lowOpenAIclosedeffort low55.7%not datedvia Epoch AI
206beam-1.0_mediumReflectionno page matchopeneffort medium55.7%not datedarcprize.org
207gpt-5-mini-2025-08-07_highOpenAIclosedeffort high54.3%not datedvia Epoch AI
208o3-2025-04-16_mediumOpenAIclosedeffort medium53.8%not datedarcprize.org
209gemini-3.5-flash-lite_highGoogle DeepMindclosedeffort high53.5%not datedarcprize.org
210thinkingmachines/Inkling-SmallEpoch: inkling-small_low, Thinking Machinesopeneffort low52.5%not datedarcprize.org
211gpt-5.4-nano-2026-03-17_xhighOpenAIclosedeffort xhigh51.5%not datedarcprize.org
212gpt-5.4-nano-2026-03-17_xhighOpenAIclosedeffort xhigh51.5%not datedarcprize.org
213claude-haiku-5-5_lowAnthropicclosedeffort low50.3%not datedarcprize.org
214gemini-3.5-flash_minimalGoogle DeepMindclosedeffort minimal48.8%not datedarcprize.org
215grok-4-fastxAIcloseddefault48.5%not datedvia Epoch AI
216thinkingmachines/Inkling-SmallEpoch: inkling-small_minimal, Thinking Machinesopeneffort minimal48.5%not datedarcprize.org
217claude-sonnet-4-5-20250929_16KAnthropiccloseddefault48.3%not datedvia Epoch AI
218claude-haiku-4-5-20251001_32KAnthropiccloseddefault47.7%not datedvia Epoch AI
219zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_low, Z.ai (Zhipu AI)openeffort low47.0%not datedarcprize.org
220claude-sonnet-4-5-20250929_8KAnthropiccloseddefault46.5%not datedvia Epoch AI
221beam-1.0_lowReflectionno page matchopeneffort low45.0%not datedarcprize.org
222zai-org/GLM-5Epoch: glm-5, Z.ai (Zhipu AI)opendefault44.7%not datedvia Epoch AI
223o3-pro-2025-06-10_lowOpenAIclosedeffort low44.3%not datedarcprize.org
224gpt-5-2025-08-07_lowOpenAIclosedeffort low44.0%not datedARC Prize Leaderboard
225o4-mini-2025-04-16_mediumOpenAIclosedeffort medium41.8%not datedarcprize.org
226o3-2025-04-16_lowOpenAIclosedeffort low41.5%not datedARC Prize Leaderboard
227gemini-2.5-pro_16KGoogle DeepMindcloseddefault41.0%not datedvia Epoch AI
228gpt-5.4-mini-2026-03-17_mediumOpenAIclosedeffort medium40.8%not datedarcprize.org
229claude-opus-4-5-20251101Anthropiccloseddefault40.0%not datedvia Epoch AI
230claude-sonnet-4-20250514_16KAnthropiccloseddefault40.0%not datedARC Prize Leaderboard
231tiny-recursion-modelnot stateddefault40.0%not datedvia Epoch AI
232gpt-5.4-nano-2026-03-17_highOpenAIclosedeffort high38.2%not datedarcprize.org
233gpt-6-luna_lowOpenAIclosedeffort low37.7%not datedarcprize.org
234claude-haiku-4-5-20251001_16KAnthropiccloseddefault37.3%not datedvia Epoch AI
235gpt-5-mini-2025-08-07_mediumOpenAIclosedeffort medium37.3%not datedvia Epoch AI
236gemini-2.5-pro_32KGoogle DeepMindcloseddefault37.0%not datedvia Epoch AI
237claude-opus-4-20250514_16KAnthropiccloseddefault35.7%not datedarcprize.org
238gemini-3.6-flash_minimalGoogle DeepMindclosedeffort minimal34.5%not datedarcprize.org
239o3-mini-2025-01-31_highOpenAIclosedeffort high34.5%not datedARC Prize Leaderboard
240gpt-5.6-luna_lowOpenAIclosedeffort low34.2%not datedarcprize.org
241Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_none, Alibabaopeneffort none34.0%not datedarcprize.org
242gemini-2.5-flash-preview-05-20Google DeepMindcloseddefault33.3%not datedarcprize.org
243gemini-2.5-flash-preview-05-20_16KGoogle DeepMindcloseddefault33.3%not datedarcprize.org
244gpt-5.1-2025-11-13_lowOpenAIclosedeffort low33.2%not datedvia Epoch AI
245gemini-2.5-pro-preview-03-25Google DeepMindcloseddefault33.0%not datedARC Prize Leaderboard
246gpt-5.4-nano-2026-03-17_mediumOpenAIclosedeffort medium33.0%not datedarcprize.org
247gemini-3.5-flash-lite_mediumGoogle DeepMindclosedeffort medium32.3%not datedarcprize.org
248gemini-2.5-flash-preview-05-20_23KGoogle DeepMindcloseddefault32.3%not datedvia Epoch AI
249gemini-2.5-flash-preview-04-17 (24K thinking)Google DeepMindcloseddefault32.3%not datedARC Prize Leaderboard
250gemini-2.5-pro-preview-06-05_1KGoogle DeepMindcloseddefault31.3%not datedARC Prize Leaderboard
251claude-sonnet-4-5-20250929_1KAnthropiccloseddefault31.0%not datedvia Epoch AI
252o1-2024-12-17_mediumOpenAIclosedeffort medium30.7%not datedARC Prize Leaderboard
253claude-opus-4-20250514_8KAnthropiccloseddefault30.7%not datedarcprize.org
254gemini-2.5-pro_8KGoogle DeepMindcloseddefault29.5%not datedvia Epoch AI
255gpt-6-sol_noneOpenAIclosedeffort none29.3%not datedarcprize.org
256claude-sonnet-4-20250514_8KAnthropiccloseddefault29.0%not datedARC Prize Leaderboard
257gemini-3-flash-previewGoogle DeepMindcloseddefault29.0%not datedARC Prize Leaderboard
258claude-3-7-sonnet-20250219_16KAnthropiccloseddefault28.6%not datedARC Prize Leaderboard
259claude-sonnet-4-20250514_1KAnthropiccloseddefault28.0%not datedARC Prize Leaderboard
260codex-mini-2025-05-16not stateddefault27.3%not datedarcprize.org
261o1-2024-12-17_lowOpenAIclosedeffort low27.2%not datedARC Prize Leaderboard
262claude-opus-4-20250514_1KAnthropiccloseddefault27.0%not datedARC Prize Leaderboard
263gpt-5-mini-2025-08-07_lowOpenAIclosedeffort low26.3%not datedvia Epoch AI
264gemini-2.5-flash-preview-05-20_8KGoogle DeepMindcloseddefault25.8%not datedarcprize.org
265claude-haiku-4-5-20251001_8KAnthropiccloseddefault25.5%not datedvia Epoch AI
266claude-sonnet-4-5-20250929Anthropiccloseddefault25.5%not datedvia Epoch AI
267claude-sonnet-4-20250514Anthropiccloseddefault23.8%not datedarcprize.org
268o1-pro-2025-03-19_lowOpenAIclosedeffort low23.3%not datedARC Prize Leaderboard
269claude-opus-4-20250514Anthropiccloseddefault22.5%not datedARC Prize Leaderboard
270o3-mini-2025-01-31_mediumOpenAIclosedeffort medium22.3%not datedarcprize.org
271gemini-3-flash-previewGoogle DeepMindcloseddefault21.5%not datedvia Epoch AI
272o4-mini-2025-04-16_lowOpenAIclosedeffort low21.3%not datedarcprize.org
273deepseek-ai/DeepSeek-R1-0528Epoch: DeepSeek-R1-0528, DeepSeekopendefault21.2%not datedarcprize.org
274claude-3-7-sonnet-20250219_8KAnthropiccloseddefault21.2%not datedARC Prize Leaderboard
275gpt-5-nano-2025-08-07_mediumOpenAIclosedeffort medium20.7%not datedvia Epoch AI
276gpt-5.4-nano-2026-03-17_lowOpenAIclosedeffort low18.3%not datedarcprize.org
277o1-preview-2024-09-12OpenAIcloseddefault18.0%not datedARC Prize Leaderboard
278gemini-3.5-flash-lite_lowGoogle DeepMindclosedeffort low17.0%not datedarcprize.org
279claude-haiku-4-5-20251001_1KAnthropiccloseddefault16.8%not datedvia Epoch AI
280gpt-5-nano-2025-08-07_highOpenAIclosedeffort high16.7%not datedvia Epoch AI
281grok-3-mini_lowxAIclosedeffort low16.5%not datedvia Epoch AI
282grok-3-mini-beta_lowxAIclosedeffort low16.5%not datedARC Prize Leaderboard
283gemini-2.5-flash-preview-05-20_1KGoogle DeepMindcloseddefault16.0%not datedARC Prize Leaderboard
284gemini-2.5-pro-preview-06-05_1KGoogle DeepMindcloseddefault16.0%not datedarcprize.org
285deepseek-ai/DeepSeek-R1Epoch: DeepSeek-R1, DeepSeekopendefault15.8%not datedARC Prize Leaderboard
286o3-mini-2025-01-31_lowOpenAIclosedeffort low14.5%not datedARC Prize Leaderboard
287claude-haiku-4-5-20251001Anthropiccloseddefault14.3%not datedvia Epoch AI
288o1-mini-2024-09-12_mediumOpenAIclosedeffort medium14.0%not datedARC Prize Leaderboard
289o1-mini-2024-09-12_unknownOpenAIclosedeffort unknown14.0%not datedvia Epoch AI
290claude-3-7-sonnet-20250219Anthropiccloseddefault13.6%not datedARC Prize Leaderboard
291deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_none, DeepSeekopeneffort none13.0%not datedarcprize.org
292gpt-5.4-mini-2026-03-17_lowOpenAIclosedeffort low13.0%not datedarcprize.org
293gpt-5.2-2025-12-11_noneOpenAIclosedeffort none12.3%not datedvia Epoch AI
294gpt-5.2-2025-12-11_unknownOpenAIclosedeffort unknown12.3%not datedarcprize.org
295deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_none, DeepSeekopeneffort none11.8%not datedarcprize.org
296claude-3-7-sonnet-20250219_1KAnthropiccloseddefault11.6%not datedARC Prize Leaderboard
297Qwen/Qwen3-235B-A22B-Instruct-2507Epoch: Qwen3-235B-A22B-Instruct-2507, Alibabaopendefault11.0%not datedvia Epoch AI
298gpt-4.5-preview-2025-02-27OpenAIcloseddefault10.3%not datedARC Prize Leaderboard
299gpt-6-luna_noneOpenAIclosedeffort none8.8%not datedarcprize.org
300deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_none, DeepSeekopeneffort none7.8%not datedarcprize.org
301gemini-3.5-flash-lite_minimalGoogle DeepMindclosedeffort minimal7.5%not datedarcprize.org
302magistral-medium-2506Mistral AIcloseddefault6.1%not datedarcprize.org
303gpt-5-2025-08-07_minimalOpenAIclosedeffort minimal6.0%not datedvia Epoch AI
304magistral-medium-2506Mistral AIcloseddefault5.9%not datedvia Epoch AI
305gpt-5.1-2025-11-13_noneOpenAIclosedeffort none5.8%not datedvia Epoch AI
306deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_none, DeepSeekopeneffort none5.7%not datedarcprize.org
307gpt-4.1-2025-04-14OpenAIcloseddefault5.5%not datedARC Prize Leaderboard
308grok-3xAIcloseddefault5.5%not datedARC Prize Leaderboard
309thinkingmachines/Inkling-SmallEpoch: inkling-small_none, Thinking Machinesopeneffort none5.5%not datedarcprize.org
310gpt-5-mini-2025-08-07_minimalOpenAIclosedeffort minimal5.3%not datedvia Epoch AI
311mistralai/Magistral-Small-2506Epoch: magistral-small-2506, Mistral AIopendefault5.0%not datedvia Epoch AI
312gpt-4o-2024-11-20OpenAIcloseddefault4.5%not datedARC Prize Leaderboard
313meta-llama/Llama-4-Maverick-17B-128E-InstructEpoch: Llama-4-Maverick-17B-128E-Instruct, Meta AIopendefault4.4%not datedarcprize.org
314gpt-5-nano-2025-08-07_lowOpenAIclosedeffort low4.0%not datedvia Epoch AI
315gpt-4.1-mini-2025-04-14OpenAIcloseddefault3.5%not datedARC Prize Leaderboard
316gpt-5-nano-2025-08-07_minimalOpenAIclosedeffort minimal1.5%not datedvia Epoch AI
317meta-llama/Llama-4-Scout-17B-16E-InstructEpoch: Llama-4-Scout-17B-16E-Instruct, Meta AIopendefault0.5%not datedARC Prize Leaderboard
318gpt-4.1-nano-2025-04-14OpenAIcloseddefault0.0%not datedARC Prize Leaderboard

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