Skip to content
Weights Not Walls
Menu

ARC-AGI-2

Reported

The second, harder set of ARC-AGI grid puzzles: 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.
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
299 in all: 53 for open-weight models, 244 for closed ones, 2 where Epoch does not say. 49 linked to a model page here.

Best open-weight result: thinkingmachines/Inkling-2-Small, 81.7%, number 53 of 299 results.

ARC-AGI-2 results
#ModelWeightsSettingScoreDateSource
1gpt-6-astra_maxOpenAIclosedeffort max95.0%not datedvia Epoch AI
2gemini-4-carbon_mediumGoogle DeepMindclosedeffort medium94.2%not datedvia Epoch AI
3gpt-6.1-sol_maxOpenAIclosedeffort max94.2%not datedvia Epoch AI
4claude-fable-5-2_xhighAnthropicclosedeffort xhigh93.8%not datedvia Epoch AI
5claude-opus-5-5_highAnthropicclosedeffort high93.3%not datedvia Epoch AI
6gpt-6-astra_xhighOpenAIclosedeffort xhigh93.3%not datedvia Epoch AI
7claude-fable-5-2_maxAnthropicclosedeffort max92.5%not datedvia Epoch AI
8claude-opus-5-5_xhighAnthropicclosedeffort xhigh92.5%not datedvia Epoch AI
9gpt-5.6-sol_maxOpenAIclosedeffort max92.5%not datedvia Epoch AI
10gemini-4-carbon_highGoogle DeepMindclosedeffort high92.1%not datedvia Epoch AI
11gpt-6-astra_highOpenAIclosedeffort high92.1%not datedvia Epoch AI
12gpt-6-astra_mediumOpenAIclosedeffort medium92.1%not datedvia Epoch AI
13claude-opus-5-5_maxAnthropicclosedeffort max91.7%not datedvia Epoch AI
14gpt-6.1-sol_highOpenAIclosedeffort high91.7%not datedvia Epoch AI
15gpt-6.1-sol_xhighOpenAIclosedeffort xhigh91.7%not datedvia Epoch AI
16gemini-4-argon_highGoogle DeepMindclosedeffort high91.3%not datedvia Epoch AI
17claude-opus-5_maxAnthropicclosedeffort max90.4%not datedvia Epoch AI
18claude-fable-5-2_highAnthropicclosedeffort high90.4%not datedvia Epoch AI
19claude-fable-5-1_maxAnthropicclosedeffort max90.0%not datedvia Epoch AI
20claude-fable-5-1_xhighAnthropicclosedeffort xhigh90.0%not datedvia Epoch AI
21gpt-5.6-sol_xhighOpenAIclosedeffort xhigh90.0%not datedvia Epoch AI
22gpt-5.6-sol_xhighOpenAIclosedeffort xhigh90.0%not datedvia Epoch AI
23claude-fable-5-2_mediumAnthropicclosedeffort medium89.6%not datedvia Epoch AI
24gpt-6-sol_maxOpenAIclosedeffort max89.6%not datedvia Epoch AI
25claude-fable-5_maxAnthropicclosedeffort max89.2%not datedvia Epoch AI
26gemini-3.8-flash_highGoogle DeepMindclosedeffort high89.2%not datedvia Epoch AI
27claude-fable-5-1_highAnthropicclosedeffort high88.8%not datedvia Epoch AI
28claude-fable-5_xhighAnthropicclosedeffort xhigh88.3%not datedvia Epoch AI
29claude-opus-5_highAnthropicclosedeffort high88.3%not datedvia Epoch AI
30claude-opus-5_maxAnthropicclosedeffort max88.3%not datedvia Epoch AI
31claude-fable-5_highAnthropicclosedeffort high87.5%not datedvia Epoch AI
32claude-opus-5-5_mediumAnthropicclosedeffort medium87.5%not datedvia Epoch AI
33gemini-4-argon_mediumGoogle DeepMindclosedeffort medium87.1%not datedvia Epoch AI
34claude-fable-5-2_lowAnthropicclosedeffort low86.7%not datedvia Epoch AI
35gpt-6.1-sol_mediumOpenAIclosedeffort medium86.7%not datedvia Epoch AI
36claude-fable-5-1_mediumAnthropicclosedeffort medium86.3%not datedvia Epoch AI
37gpt-5.6-sol_highOpenAIclosedeffort high85.4%not datedvia Epoch AI
38gpt-6-astra_lowOpenAIclosedeffort low85.4%not datedvia Epoch AI
39gpt-5.5_xhighOpenAIclosedeffort xhigh85.0%not datedvia Epoch AI
40gpt-5.5_xhighOpenAIclosedeffort xhigh85.0%not datedvia Epoch AI
41gemini-3.7-flash_highGoogle DeepMindclosedeffort high84.6%not datedvia Epoch AI
42gemini-3-deep-think-previewGoogle DeepMindcloseddefault84.6%not datedvia Epoch AI
43gpt-5.5-pro_highOpenAIclosedeffort high84.6%not datedvia Epoch AI
44gpt-5.5-pro_xhighOpenAIclosedeffort xhigh84.2%not datedvia Epoch AI
45gpt-5.5-pro_xhighOpenAIclosedeffort xhigh84.2%not datedvia Epoch AI
46gpt-5.6-terra_maxOpenAIclosedeffort max83.9%not datedvia Epoch AI
47claude-sonnet-5-5_xhighAnthropicclosedeffort xhigh83.9%not datedvia Epoch AI
48gpt-5.4-pro-2026-03-05_xhighOpenAIclosedeffort xhigh83.3%not datedvia Epoch AI
49gpt-5.4-pro-2026-03-05_xhighOpenAIclosedeffort xhigh83.3%not datedvia Epoch AI
50gpt-5.5_highOpenAIclosedeffort high83.3%not datedvia Epoch AI
51gemini-3.8-flash_mediumGoogle DeepMindclosedeffort medium82.9%not datedvia Epoch AI
52claude-fable-5_mediumAnthropicclosedeffort medium82.5%not datedvia Epoch AI
53thinkingmachines/Inkling-2-SmallEpoch: inkling-2-small_xhigh, Thinking Machinesopeneffort xhigh81.7%not datedvia Epoch AI
54claude-fable-5-1_lowAnthropicclosedeffort low78.3%not datedvia Epoch AI
55deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_max, DeepSeekopeneffort max78.1%not datedvia Epoch AI
56gpt-6-sol_xhighOpenAIclosedeffort xhigh78.1%not datedvia Epoch AI
57gemini-3.8-flash_lowGoogle DeepMindclosedeffort low77.5%not datedvia Epoch AI
58gemini-3.1-pro-previewGoogle DeepMindcloseddefault77.1%not datedvia Epoch AI
59gemini-3.1-pro-previewGoogle DeepMindcloseddefault77.1%not datedvia Epoch AI
60claude-fable-5_lowAnthropicclosedeffort low76.8%not datedvia Epoch AI
61gpt-6.1-sol_lowOpenAIclosedeffort low76.7%not datedvia Epoch AI
62claude-opus-4-7_maxAnthropicclosedeffort max75.8%not datedvia Epoch AI
63claude-opus-4-7_maxAnthropicclosedeffort max75.8%not datedvia Epoch AI
64gpt-5.6-terra_xhighOpenAIclosedeffort xhigh74.2%not datedvia Epoch AI
65gpt-5.6-terra_xhighOpenAIclosedeffort xhigh74.2%not datedvia Epoch AI
66gpt-5.4-2026-03-05_xhighOpenAIclosedeffort xhigh74.0%not datedvia Epoch AI
67gpt-5.4-2026-03-05_xhighOpenAIclosedeffort xhigh74.0%not datedvia Epoch AI
68deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_low, DeepSeekopeneffort low73.6%not datedvia Epoch AI
69deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_max, DeepSeekopeneffort max72.9%not datedvia Epoch AI
70claude-opus-4-8_highAnthropicclosedeffort high72.1%not datedvia Epoch AI
71gemini-3.5-flash_highGoogle DeepMindclosedeffort high72.1%not datedvia Epoch AI
72claude-opus-4-8_mediumAnthropicclosedeffort medium71.7%not datedvia Epoch AI
73gpt-5.5_mediumOpenAIclosedeffort medium70.4%not datedvia Epoch AI
74claude-opus-5-5_lowAnthropicclosedeffort low70.1%not datedvia Epoch AI
75claude-opus-4-6_120KAnthropiccloseddefault69.2%not datedvia Epoch AI
76gpt-6-sol_highOpenAIclosedeffort high68.9%not datedvia Epoch AI
77claude-opus-4-6_120KAnthropiccloseddefault68.8%not datedvia Epoch AI
78deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_high, DeepSeekopeneffort high68.6%not datedvia Epoch AI
79claude-opus-4-7_highAnthropicclosedeffort high68.3%not datedvia Epoch AI
80claude-opus-4-7_highAnthropicclosedeffort high68.3%not datedvia Epoch AI
81claude-opus-4-7_mediumAnthropicclosedeffort medium67.5%not datedvia Epoch AI
82deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_high, DeepSeekopeneffort high67.5%not datedvia Epoch AI
83gpt-5.4-2026-03-05_highOpenAIclosedeffort high67.5%not datedvia Epoch AI
84claude-sonnet-5-5_highAnthropicclosedeffort high67.4%not datedvia Epoch AI
85grok-4.6_xhighxAIclosedeffort xhigh67.1%not datedvia Epoch AI
86gpt-5.6-sol_mediumOpenAIclosedeffort medium67.1%not datedvia Epoch AI
87gpt-5.6-terra_highOpenAIclosedeffort high67.1%not datedvia Epoch AI
88claude-opus-4-6_120KAnthropiccloseddefault66.3%not datedvia Epoch AI
89zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_max, Z.ai (Zhipu AI)openeffort max65.8%not datedvia Epoch AI
90grok-4-20xAIcloseddefault65.1%not datedvia Epoch AI
91grok-4.6_highxAIclosedeffort high65.1%not datedvia Epoch AI
92claude-opus-4-6_120KAnthropiccloseddefault64.6%not datedvia Epoch AI
93gemini-3.7-flash_mediumGoogle DeepMindclosedeffort medium63.7%not datedvia Epoch AI
94claude-opus-4-8_lowAnthropicclosedeffort low62.2%not datedvia Epoch AI
95claude-opus-4-7_lowAnthropicclosedeffort low62.1%not datedvia Epoch AI
96deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_max, DeepSeekopeneffort max61.4%not datedvia Epoch AI
97grok-4.7_xhighxAIclosedeffort xhigh61.4%not datedvia Epoch AI
98claude-haiku-5-5_xhighAnthropicclosedeffort xhigh61.3%not datedvia Epoch AI
99deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_max, DeepSeekopeneffort max61.3%not datedvia Epoch AI
100grok-4.6_mediumxAIclosedeffort medium61.3%not datedvia Epoch AI
101deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_low, DeepSeekopeneffort low60.6%not datedvia Epoch AI
102claude-sonnet-4-6_highAnthropicclosedeffort high60.4%not datedvia Epoch AI
103gemini-3.6-flash_highGoogle DeepMindclosedeffort high60.4%not datedvia Epoch AI
104moonshotai/Kimi-K3Epoch: kimi-k3_max, Moonshotopeneffort max60.4%not datedvia Epoch AI
105deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_high, DeepSeekopeneffort high59.7%not datedvia Epoch AI
106gpt-6-astra_noneOpenAIclosedeffort none59.6%not datedvia Epoch AI
107gpt-5.6-luna_maxOpenAIclosedeffort max59.5%not datedvia Epoch AI
108gpt-6-luna_maxOpenAIclosedeffort max59.3%not datedvia Epoch AI
109grok-4.7_mediumxAIclosedeffort medium58.8%not datedvia Epoch AI
110grok-4.7_highxAIclosedeffort high58.3%not datedvia Epoch AI
111claude-sonnet-4-6_maxAnthropicclosedeffort max58.3%not datedvia Epoch AI
112gpt-6-sol_mediumOpenAIclosedeffort medium57.8%not datedvia Epoch AI
113meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_xhigh, Meta AIopeneffort xhigh57.5%not datedvia Epoch AI
114deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_low, DeepSeekopeneffort low56.3%not datedvia Epoch AI
115deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_high, DeepSeekopeneffort high56.0%not datedvia Epoch AI
116gpt-5.4-2026-03-05_mediumOpenAIclosedeffort medium55.4%not datedvia Epoch AI
117moonshotai/Kimi-K3Epoch: kimi-k3_high, Moonshotopeneffort high55.0%not datedvia Epoch AI
118gpt-5.2-pro-2025-12-11_highOpenAIclosedeffort high54.2%not datedvia Epoch AI
119gemini-3.7-flash_lowGoogle DeepMindclosedeffort low52.9%not datedvia Epoch AI
120gpt-5.2-2025-12-11_xhighOpenAIclosedeffort xhigh52.9%not datedvia Epoch AI
121gpt-5.2-2025-12-11_xhighOpenAIclosedeffort xhigh52.9%not datedvia Epoch AI
122gpt-5.2-2025-12-11_xhighOpenAIclosedeffort xhigh52.9%not datedvia Epoch AI
123grok-4.5_highxAIclosedeffort high52.6%not datedvia Epoch AI
124grok-4.5_mediumxAIclosedeffort medium52.6%not datedvia Epoch AI
125gemini-3.6-flash_mediumGoogle DeepMindclosedeffort medium50.4%not datedvia Epoch AI
126zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_high, Z.ai (Zhipu AI)openeffort high50.1%not datedvia Epoch AI
127gemini-4-carbon_lowGoogle DeepMindclosedeffort low49.6%not datedvia Epoch AI
128gpt-5.6-luna_xhighOpenAIclosedeffort xhigh47.6%not datedvia Epoch AI
129gpt-5.6-luna_xhighOpenAIclosedeffort xhigh47.6%not datedvia Epoch AI
130meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_high, Meta AIopeneffort high47.5%not datedvia Epoch AI
131deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_low, DeepSeekopeneffort low46.0%not datedvia Epoch AI
132gemini-3-deep-think-previewGoogle DeepMindcloseddefault45.1%not datedvia Epoch AI
133gpt-5.2-2025-12-11_highOpenAIclosedeffort high43.3%not datedvia Epoch AI
134gpt-5.6-sol_lowOpenAIclosedeffort low42.5%not datedvia Epoch AI
135Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_xhigh, Alibabaopeneffort xhigh42.4%not datedvia Epoch AI
136gemini-4-argon_lowGoogle DeepMindclosedeffort low42.1%not datedvia Epoch AI
137gpt-6-luna_xhighOpenAIclosedeffort xhigh41.9%not datedvia Epoch AI
138thinkingmachines/Inkling-SmallEpoch: inkling-small_xhigh, Thinking Machinesopeneffort xhigh40.1%not datedvia Epoch AI
139gpt-5.2-pro-2025-12-11_mediumOpenAIclosedeffort medium38.5%not datedvia Epoch AI
140claude-opus-4-5-20251101_64KAnthropiccloseddefault37.6%not datedvia Epoch AI
141gpt-5.6-terra_mediumOpenAIclosedeffort medium37.5%not datedvia Epoch AI
142thinkingmachines/InklingEpoch: Inkling, Thinking Machinesopendefault36.5%not datedvia Epoch AI
143gemini-3-flash-previewGoogle DeepMindcloseddefault33.6%not datedvia Epoch AI
144gpt-5.5_lowOpenAIclosedeffort low33.3%not datedvia Epoch AI
145grok-4.5_lowxAIclosedeffort low33.1%not datedvia Epoch AI
146thinkingmachines/Inkling-SmallEpoch: inkling-small_high, Thinking Machinesopeneffort high33.1%not datedvia Epoch AI
147claude-sonnet-5-5_mediumAnthropicclosedeffort medium32.8%not datedvia Epoch AI
148gpt-6-sol_lowOpenAIclosedeffort low31.5%not datedvia Epoch AI
149gpt-6-luna_highOpenAIclosedeffort high31.4%not datedvia Epoch AI
150gemini-3-pro-previewGoogle DeepMindcloseddefault31.1%not datedvia Epoch AI
151claude-haiku-5-5_highAnthropicclosedeffort high30.7%not datedvia Epoch AI
152gemini-3.6-flash_lowGoogle DeepMindclosedeffort low30.4%not datedvia Epoch AI
153gpt-5.6-luna_highOpenAIclosedeffort high29.3%not datedvia Epoch AI
154gpt-5.4-2026-03-05_lowOpenAIclosedeffort low29.2%not datedvia Epoch AI
155zai-org/GLM-5.3-FlashEpoch: glm-5.3-flash_low, Z.ai (Zhipu AI)openeffort low27.9%not datedvia Epoch AI
156grok-4.6_lowxAIclosedeffort low27.6%not datedvia Epoch AI
157meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_medium, Meta AIopeneffort medium27.5%not datedvia Epoch AI
158gpt-5.2-2025-12-11_mediumOpenAIclosedeffort medium26.7%not datedvia Epoch AI
159claude-sonnet-5-5_lowAnthropicclosedeffort low24.0%not datedvia Epoch AI
160beam-1.0_xhighReflectionno page matchopeneffort xhigh23.5%not datedvia Epoch AI
161claude-opus-4-5-20251101_16KAnthropiccloseddefault22.8%not datedvia Epoch AI
162Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_low, Alibabaopeneffort low22.8%not datedvia Epoch AI
163zai-org/GLM-5.2Epoch: glm-5.2_unknown, Z.ai (Zhipu AI)openeffort unknown22.8%not datedvia Epoch AI
164beam-1.0_highReflectionno page matchopeneffort high20.0%not datedvia Epoch AI
165gpt-5.4-mini-2026-03-17_xhighOpenAIclosedeffort xhigh18.9%not datedvia Epoch AI
166gpt-5.4-mini-2026-03-17_xhighOpenAIclosedeffort xhigh18.9%not datedvia Epoch AI
167gpt-5.6-terra_lowOpenAIclosedeffort low18.8%not datedvia Epoch AI
168gpt-5-pro-2025-10-06_unknownOpenAIclosedeffort unknown18.3%not datedvia Epoch AI
169gpt-6-luna_mediumOpenAIclosedeffort medium18.1%not datedvia Epoch AI
170gpt-5.1-2025-11-13_highOpenAIclosedeffort high17.6%not datedvia Epoch AI
171grok-4.7_lowxAIclosedeffort low16.7%not datedvia Epoch AI
172grok-4-0709xAIcloseddefault16.0%not datedvia Epoch AI
173claude-opus-4-5-20251101_8KAnthropiccloseddefault13.9%not datedvia Epoch AI
174thinkingmachines/Inkling-SmallEpoch: inkling-small_medium, Thinking Machinesopeneffort medium13.6%not datedvia Epoch AI
175claude-sonnet-4-5-20250929_32KAnthropiccloseddefault13.6%not datedvia Epoch AI
176Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_medium, Alibabaopeneffort medium13.2%not datedvia Epoch AI
177gpt-5.4-mini-2026-03-17_highOpenAIclosedeffort high13.2%not datedvia Epoch AI
178gemini-3-flash-previewGoogle DeepMindcloseddefault12.8%not datedvia Epoch AI
179moonshotai/Kimi-K3Epoch: kimi-k3_low, Moonshotopeneffort low12.4%not datedvia Epoch AI
180meta-models/Muse-Spark-Open-816B-A42BEpoch: Muse-Spark-Open-816B-A42B_low, Meta AIopeneffort low12.2%not datedvia Epoch AI
181moonshotai/Kimi-K2.5Epoch: kimi-k2.5, Moonshotopendefault11.8%not datedvia Epoch AI
182gemini-3.5-flash-lite_highGoogle DeepMindclosedeffort high10.3%not datedvia Epoch AI
183gpt-5-2025-08-07_highOpenAIclosedeffort high9.9%not datedvia Epoch AI
184gpt-5.2-2025-12-11_lowOpenAIclosedeffort low9.7%not datedvia Epoch AI
185claude-haiku-5-5_mediumAnthropicclosedeffort medium9.4%not datedvia Epoch AI
186gemini-3.5-flash_minimalGoogle DeepMindclosedeffort minimal8.9%not datedvia Epoch AI
187claude-opus-4-20250514_16KAnthropiccloseddefault8.6%not datedvia Epoch AI
188claude-opus-4-5-20251101Anthropiccloseddefault7.8%not datedvia Epoch AI
189gpt-5-2025-08-07_mediumOpenAIclosedeffort medium7.5%not datedvia Epoch AI
190gpt-5.6-luna_mediumOpenAIclosedeffort medium7.4%not datedvia Epoch AI
191claude-sonnet-4-5-20250929_16KAnthropiccloseddefault6.9%not datedvia Epoch AI
192claude-sonnet-4-5-20250929_8KAnthropiccloseddefault6.9%not datedvia Epoch AI
193gpt-5.1-2025-11-13_mediumOpenAIclosedeffort medium6.5%not datedvia Epoch AI
194o3-2025-04-16_highOpenAIclosedeffort high6.5%not datedvia Epoch AI
195beam-1.0_mediumReflectionno page matchopeneffort medium6.4%not datedvia Epoch AI
196tiny-recursion-modelnot stateddefault6.3%not datedvia Epoch AI
197o4-mini-2025-04-16_highOpenAIclosedeffort high6.1%not datedvia Epoch AI
198claude-sonnet-4-20250514_16KAnthropiccloseddefault5.9%not datedvia Epoch AI
199claude-sonnet-4-5-20250929_1KAnthropiccloseddefault5.8%not datedvia Epoch AI
200gpt-5.4-nano-2026-03-17_xhighOpenAIclosedeffort xhigh5.7%not datedvia Epoch AI
201gpt-5.4-nano-2026-03-17_xhighOpenAIclosedeffort xhigh5.7%not datedvia Epoch AI
202grok-4-fastxAIcloseddefault5.3%not datedvia Epoch AI
203gemini-3.5-flash-lite_mediumGoogle DeepMindclosedeffort medium5.3%not datedvia Epoch AI
204gpt-5.6-luna_lowOpenAIclosedeffort low5.1%not datedvia Epoch AI
205thinkingmachines/Inkling-SmallEpoch: inkling-small_low, Thinking Machinesopeneffort low4.9%not datedvia Epoch AI
206gemini-2.5-pro_32KGoogle DeepMindcloseddefault4.9%not datedvia Epoch AI
207zai-org/GLM-5Epoch: glm-5, Z.ai (Zhipu AI)opendefault4.9%not datedvia Epoch AI
208MiniMaxAI/MiniMax-M2.5Epoch: MiniMax-M2.5, MiniMaxopendefault4.9%not datedvia Epoch AI
209o3-pro-2025-06-10_highOpenAIclosedeffort high4.9%not datedvia Epoch AI
210gpt-6-luna_lowOpenAIclosedeffort low4.6%not datedvia Epoch AI
211claude-opus-4-20250514_8KAnthropiccloseddefault4.5%not datedvia Epoch AI
212gpt-5-mini-2025-08-07_highOpenAIclosedeffort high4.4%not datedvia Epoch AI
213gpt-5.4-mini-2026-03-17_mediumOpenAIclosedeffort medium4.4%not datedvia Epoch AI
214claude-haiku-4-5-20251001_32KAnthropiccloseddefault4.0%not datedvia Epoch AI
215deepseek-ai/DeepSeek-V3.2Epoch: deepseek/deepseek-v3.2, DeepSeekopendefault4.0%not datedvia Epoch AI
216gemini-2.5-pro_16KGoogle DeepMindcloseddefault4.0%not datedvia Epoch AI
217gpt-5-mini-2025-08-07_mediumOpenAIclosedeffort medium4.0%not datedvia Epoch AI
218thinkingmachines/Inkling-SmallEpoch: inkling-small_minimal, Thinking Machinesopeneffort minimal4.0%not datedvia Epoch AI
219claude-sonnet-4-5-20250929Anthropiccloseddefault3.8%not datedvia Epoch AI
220gpt-5.4-nano-2026-03-17_highOpenAIclosedeffort high3.6%not datedvia Epoch AI
221gemini-3-flash-previewGoogle DeepMindcloseddefault3.3%not datedvia Epoch AI
222o3-mini-2025-01-31_highOpenAIclosedeffort high3.0%not datedvia Epoch AI
223o3-2025-04-16_mediumOpenAIclosedeffort medium3.0%not datedvia Epoch AI
224gemini-2.5-pro_8KGoogle DeepMindcloseddefault2.9%not datedvia Epoch AI
225claude-haiku-4-5-20251001_16KAnthropiccloseddefault2.8%not datedvia Epoch AI
226beam-1.0_lowReflectionno page matchopeneffort low2.8%not datedvia Epoch AI
227claude-haiku-5-5_lowAnthropicclosedeffort low2.8%not datedvia Epoch AI
228gemini-3.6-flash_minimalGoogle DeepMindclosedeffort minimal2.6%not datedvia Epoch AI
229gpt-5-nano-2025-08-07_highOpenAIclosedeffort high2.6%not datedvia Epoch AI
230gemini-2.5-flash-preview-05-20_23KGoogle DeepMindcloseddefault2.5%not datedvia Epoch AI
231o4-mini-2025-04-16_mediumOpenAIclosedeffort medium2.4%not datedvia Epoch AI
232gemini-2.5-flash-preview-05-20_1KGoogle DeepMindcloseddefault2.2%not datedvia Epoch AI
233claude-sonnet-4-20250514_8KAnthropiccloseddefault2.1%not datedvia Epoch AI
234gemini-2.5-flash-preview-05-20_8KGoogle DeepMindcloseddefault2.1%not datedvia Epoch AI
235deepseek-ai/DeepSeek-V4-Flash-0731Epoch: deepseek-v4-flash-0731_none, DeepSeekopeneffort none2.1%not datedvia Epoch AI
236deepseek-ai/DeepSeek-V4.1-Flash-1017Epoch: deepseek-v4.1-flash-1017_none, DeepSeekopeneffort none2.1%not datedvia Epoch AI
237o3-mini-2025-01-31_mediumOpenAIclosedeffort medium2.1%not datedvia Epoch AI
238o3-pro-2025-06-10_lowOpenAIclosedeffort low2.1%not datedvia Epoch AI
239o3-2025-04-16_lowOpenAIclosedeffort low2.0%not datedvia Epoch AI
240gemini-2.5-flash-preview-05-20_16KGoogle DeepMindcloseddefault2.0%not datedvia Epoch AI
241gpt-5-2025-08-07_lowOpenAIclosedeffort low1.9%not datedvia Epoch AI
242gpt-5-2025-08-07_lowOpenAIclosedeffort low1.9%not datedvia Epoch AI
243gpt-5.1-2025-11-13_lowOpenAIclosedeffort low1.9%not datedvia Epoch AI
244gpt-5.4-nano-2026-03-17_mediumOpenAIclosedeffort medium1.9%not datedvia Epoch AI
245o3-pro-2025-06-10_mediumOpenAIclosedeffort medium1.9%not datedvia Epoch AI
246gemini-2.5-flash-preview-05-20Google DeepMindcloseddefault1.7%not datedvia Epoch AI
247claude-haiku-4-5-20251001_8KAnthropiccloseddefault1.7%not datedvia Epoch AI
248gpt-5-mini-2025-08-07_minimalOpenAIclosedeffort minimal1.7%not datedvia Epoch AI
249o4-mini-2025-04-16_lowOpenAIclosedeffort low1.7%not datedvia Epoch AI
250gpt-6-sol_noneOpenAIclosedeffort none1.7%not datedvia Epoch AI
251gpt-5.4-nano-2026-03-17_lowOpenAIclosedeffort low1.5%not datedvia Epoch AI
252gemini-3.5-flash-lite_lowGoogle DeepMindclosedeffort low1.5%not datedvia Epoch AI
253Qwen/Qwen3.8-27BEpoch: qwen3.8-27b_none, Alibabaopeneffort none1.5%not datedvia Epoch AI
254deepseek-ai/DeepSeek-R1Epoch: DeepSeek-R1, DeepSeekopendefault1.3%not datedvia Epoch AI
255gemini-2.0-flash-001Google DeepMind,Googlecloseddefault1.3%not datedvia Epoch AI
256claude-opus-4-20250514Anthropiccloseddefault1.3%not datedvia Epoch AI
257claude-sonnet-4-20250514Anthropiccloseddefault1.3%not datedvia Epoch AI
258codex-mini-2025-05-16not stateddefault1.3%not datedvia Epoch AI
259claude-haiku-4-5-20251001Anthropiccloseddefault1.3%not datedvia Epoch AI
260claude-haiku-4-5-20251001_1KAnthropiccloseddefault1.3%not datedvia Epoch AI
261deepseek-ai/DeepSeek-V4.1-FlashEpoch: deepseek-v4.1-flash_none, DeepSeekopeneffort none1.3%not datedvia Epoch AI
262gemini-3-flash-previewGoogle DeepMindcloseddefault1.3%not datedvia Epoch AI
263Qwen/Qwen3-235B-A22B-Instruct-2507Epoch: Qwen3-235B-A22B-Instruct-2507, Alibabaopendefault1.3%not datedvia Epoch AI
264deepseek-ai/DeepSeek-R1-0528Epoch: DeepSeek-R1-0528, DeepSeekopendefault1.1%not datedvia Epoch AI
265gpt-5.4-mini-2026-03-17_lowOpenAIclosedeffort low1.1%not datedvia Epoch AI
266claude-3-7-sonnet-20250219_8KAnthropiccloseddefault0.9%not datedvia Epoch AI
267gpt-5-nano-2025-08-07_mediumOpenAIclosedeffort medium0.9%not datedvia Epoch AI
268claude-sonnet-4-20250514_1KAnthropiccloseddefault0.9%not datedvia Epoch AI
269deepseek-ai/DeepSeek-V4-Pro-0813Epoch: deepseek-v4-pro-0813_none, DeepSeekopeneffort none0.8%not datedvia Epoch AI
270gemini-3.5-flash-lite_minimalGoogle DeepMindclosedeffort minimal0.8%not datedvia Epoch AI
271gpt-5-mini-2025-08-07_lowOpenAIclosedeffort low0.8%not datedvia Epoch AI
272gpt-5.2-2025-12-11_unknownOpenAIclosedeffort unknown0.8%not datedvia Epoch AI
273o1-mini-2024-09-12_unknownOpenAIclosedeffort unknown0.8%not datedvia Epoch AI
274gemini-1.5-pro-002Google DeepMindcloseddefault0.8%not datedvia Epoch AI
275gpt-4.5-preview-2025-02-27OpenAIcloseddefault0.8%not datedvia Epoch AI
276claude-3-7-sonnet-20250219_16KAnthropiccloseddefault0.7%not datedvia Epoch AI
277gpt-4.1-2025-04-14OpenAIcloseddefault0.4%not datedvia Epoch AI
278gpt-5.1-2025-11-13_noneOpenAIclosedeffort none0.4%not datedvia Epoch AI
279grok-3-mini_lowxAIclosedeffort low0.4%not datedvia Epoch AI
280claude-3-7-sonnet-20250219_1KAnthropiccloseddefault0.4%not datedvia Epoch AI
281claude-3-7-sonnet-20250219Anthropiccloseddefault0.0%not datedvia Epoch AI
282claude-opus-4-20250514_1KAnthropiccloseddefault0.0%not datedvia Epoch AI
283gemini-2.5-pro-preview-06-05_1KGoogle DeepMindcloseddefault0.0%not datedvia Epoch AI
284gpt-4.1-mini-2025-04-14OpenAIcloseddefault0.0%not datedvia Epoch AI
285gpt-4.1-nano-2025-04-14OpenAIcloseddefault0.0%not datedvia Epoch AI
286gpt-4o-2024-11-20OpenAIcloseddefault0.0%not datedvia Epoch AI
287gpt-4o-mini-2024-07-18OpenAIcloseddefault0.0%not datedvia Epoch AI
288gpt-5-2025-08-07_minimalOpenAIclosedeffort minimal0.0%not datedvia Epoch AI
289gpt-5-nano-2025-08-07_lowOpenAIclosedeffort low0.0%not datedvia Epoch AI
290gpt-5-nano-2025-08-07_minimalOpenAIclosedeffort minimal0.0%not datedvia Epoch AI
291gpt-6-luna_noneOpenAIclosedeffort none0.0%not datedvia Epoch AI
292grok-3xAIcloseddefault0.0%not datedvia Epoch AI
293thinkingmachines/Inkling-SmallEpoch: inkling-small_none, Thinking Machinesopeneffort none0.0%not datedvia Epoch AI
294meta-llama/Llama-4-Maverick-17B-128E-InstructEpoch: Llama-4-Maverick-17B-128E-Instruct, Meta AIopendefault0.0%not datedvia Epoch AI
295meta-llama/Llama-4-Scout-17B-16E-InstructEpoch: Llama-4-Scout-17B-16E-Instruct, Meta AIopendefault0.0%not datedvia Epoch AI
296magistral-medium-2506Mistral AIcloseddefault0.0%not datedvia Epoch AI
297magistral-medium-2506Mistral AIcloseddefault0.0%not datedvia Epoch AI
298mistralai/Magistral-Small-2506Epoch: magistral-small-2506, Mistral AIopendefault0.0%not datedvia Epoch AI
299o3-mini-2025-01-31_lowOpenAIclosedeffort low0.0%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