Inference
ARC Prize Foundation

ARC-AGI-1 (semi-private)

Abstract reasoning puzzles held out from training data.

ModelsModels with a result.109
Top resultBest result on this test.98.5%Claude Fable 5
Top-3 spreadPoints from first to third.1.0 pts
YearYear of release.N/A

Result and cost

020406080100$0.001$0.003$0.01$0.03$0.1$0.3$1$3$10$30$100RESULTUS DOLLARS PER TASK · LOG SCALE

Results

109 results
#ModelResultThe score from the source.CostUS dollars to run one task.
01Claude Fable 5max effort98.5%$2.11
02Gemini 3.1 Pro Preview98.0%$0.522
03GPT-6 Astramax effort97.5%$0.433
03Claude Fable 5.1max effort97.5%$1.4
03Claude Opus 5high effort97.5%$0.47
06GPT-5.6 Solmax effort96.5%$0.54
06GPT-5.6 Terramax effort96.5%$0.55
08Gemini 3 Deep Think (2/26)96.0%$7.17
09Gemini 3.7 Flashhigh effort95.5%$0.12
10GPT-5.5 Proxhigh effort95.0%$4.52
10GPT-5.5xhigh effort95.0%$0.73
12GPT-5.4 Proxhigh effort94.5%$7.21
12GPT-5.2 (Refine.)94.5%$11.4
14Claude Opus 4.6 (120K, High)94.0%$1.39
15GPT-5.4xhigh effort93.7%$0.62
16Claude Opus 4.6 (120K, Max)93.0%$1.88
17Claude Opus 4.8max effort92.5%$2.33
17Gemini 3.5 Flashhigh effort92.5%$0.428
19Claude 4.7max effort92.0%$2.58
19Claude Opus 4.6 (120K, Medium)92.0%$0.965
21Gemini 3.6 Flashhigh effort91.2%$0.336
22GPT-5.6 Luna 2026-07-30max effort90.7%$0.065
23GPT-5.2 Proxhigh effort90.5%$11.65
24Grok 4.20 (Reasoning)89.5%$0.35
25GPT-5.6 Lunamax effort88.0%$0.32
26Gemini 3 Deep Think (Preview) ²87.5%$44.26
27Grok 4.6xhigh effort87.0%$0.348
28GPT-5.2xhigh effort86.2%$0.96
29Claude Sonnet 4.6max effort86.0%$1.45
29Claude Opus 4.6 (120K, Low)86.0%$0.648
31Grok 4.5high effort85.7%$0.33
32Gemini 3 Flash Previewhigh effort84.7%$0.174
33Inkling-Smallxhigh effort84.0%$0.115
34Opus 4.5 (Thinking, 64K)80.0%$1.47
35Grok 4 (Refine.)79.6%$8.42
36Inkling79.5%$0.3
37GLM-5.277.0%$0.192
38Opus 4.5 (Thinking, 32K)75.8%$0.95
39Gemini 3 Pro75.0%$0.493
40GPT-5.1 (Thinking, High)72.8%$0.674
41Opus 4.5 (Thinking, 16K)72.0%$0.62
42GPT-5 Pro70.2%$4.78
43Grok 4 (Thinking)66.7%$1.01
44GPT-5high effort65.7%$0.509
45Kimi K2.565.3%$0.15
46GPT-5.4 minixhigh effort63.7%$0.51
46MiniMax M2.563.7%$0.07
46Claude Sonnet 4.5 (Thinking 32K)63.7%$0.519
49o3high effort60.8%$0.5
50o3-prohigh effort59.3%$4.16
51o4-mini (high)high effort58.7%$0.406
51Opus 4.5 (Thinking, 8K)58.7%$0.38
53GPT-5.1 (Thinking, Medium)57.7%$0.315
54DeepSeek V3.257.0%$0.08
55GPT-5 minihigh effort54.3%$0.116
56Gemini 3.5 Flash-Litehigh effort53.5%$0.093
57GPT-5.4 nanoxhigh effort51.5%$0.13
58Grok 4 Fast (Reasoning)48.5%$0.031
59Claude Sonnet 4.5 (Thinking 16K)48.3%$0.268
60Claude Haiku 4.5 (Thinking 32K)47.7%$0.256
61Claude Sonnet 4.5 (Thinking 8K)46.5%$0.178
62GLM-544.7%$0.17
63Gemini 2.5 Pro (Thinking 16K)41.0%$0.484
64Claude Sonnet 4 (Thinking 16K)40.0%$0.366
64Opus 4.5 (Thinking, None)40.0%$0.13
64Tiny Recursion Model (TRM)40.0%$1.76
67Claude Haiku 4.5 (Thinking 16K)37.3%$0.104
68Gemini 2.5 Pro (Thinking 32K)37.0%$0.512
69Claude Opus 4 (Thinking 16K)35.7%$1.25
70o3-minihigh effort34.5%$0.399
71Gemini 2.5 Flash (Preview)33.3%$0.037
71Gemini 2.5 Flash (Preview) (Thinking 16K)33.3%$0.213
73GPT-5.1 (Thinking, Low)33.2%$0.1
74Gemini 2.5 Flash (Preview) (Thinking 24K)32.3%$0.197
75Claude Sonnet 4.5 (Thinking 1K)31.0%$0.093
76Claude Opus 4 (Thinking 8K)30.7%$0.741
77Gemini 2.5 Pro (Thinking 8K)29.5%$0.295
78Claude Sonnet 4 (Thinking 8K)29.0%$0.195
79Claude Sonnet 4 (Thinking 1K)28.0%$0.094
80Codex Mini (Latest)27.3%$0.16
81Claude Opus 4 (Thinking 1K)27.0%$0.502
82Gemini 2.5 Flash (Preview) (Thinking 8K)25.8%$0.134
83Claude Sonnet 4.525.5%$0.081
83Claude Haiku 4.5 (Thinking 8K)25.5%$0.069
85Claude Sonnet 423.8%$0.081
86Claude Opus 422.5%$0.404
87Deepseek R1 (05/28)21.2%$0.046
88Claude 3.78k effort21.2%$0.21
89Claude Haiku 4.5 (Thinking 1K)16.8%$0.031
90GPT-5 nanohigh effort16.7%$0.029
91Grok 3 Minilow effort16.5%<$0.01
92Gemini 2.5 Flash (Preview) (Thinking 1K)16.0%$0.036
92Gemini 2.5 Pro (Thinking 1K)16.0%$0.057
94DeepSeek-R115.8%$0.06
95Claude Haiku 4.514.3%$0.026
96O1 Mini14.0%$0.135
97Qwen3-235b-a22b Instruct (25/07)11.0%<$0.01
98GPT-4.510.3%$0.29
99Magistral Medium (Thinking)6.1%$0.099
100Magistral Medium5.9%$0.102
101GPT-5.1 (Thinking, None)5.8%$0.034
102Grok 3 [Beta]5.5%$0.093
102GPT-4.15.5%$0.039
104Magistral Small5.0%$0.04
105GPT-4o4.5%$0.05
106Llama 4 Maverick4.4%<$0.01
107GPT-4.1 mini3.5%<$0.01
108Llama 4 Scout0.5%<$0.01
109GPT-4.1 nano0.0%<$0.01