ARC-AGI-3

Interactive turn-based games with unstated rules that the model must explore, learn and win, semi-private set, scored by the ARC Prize Foundation. The standard harness is shown without a harness label; runs through a model provider's own adapter are labelled Provider adapter.

CategoryReasoning
MetricSemi-private games score
SourceARC Prize (original)
LicenseInternal use
Benchmark pagearcprize.org
StatusScored
Weight1.2 within Reasoning
FitDiscrimination 4.16, noise 2.44
LineageShares weight with arc_agi_3

Results

#ModelProviderResultDate
1 GPT-6 Astra High OpenAI 100.0% 2026-11-11
2 Gemini 4 Carbon Medium Google 99.1% 2026-11-11
3 GPT-6 Astra Max OpenAI 98.6% 2026-11-11
4 GPT-6 Astra Medium OpenAI 98.4% 2026-11-11
5 GPT-6 Astra Extra High OpenAI 98.4% 2026-11-11
6 GPT-6 Astra Low OpenAI 98.0% 2026-11-11
7 Claude Fable 5.2 High Anthropic 97.7% 2026-11-11
8 Gemini 4 Carbon Low Google 97.7% 2026-11-11
9 Gemini 4 Argon Low Google 97.5% 2026-11-11
10 Gemini 4 Argon Medium Google 97.3% 2026-11-11
11 Gemini 4 Argon High Google 97.3% 2026-11-11
12 Claude Fable 5.2 Extra High Anthropic 97.3% 2026-11-11
13 Gemini 4 Carbon High Google 96.9% 2026-11-11
14 Claude Fable 5.2 Medium Anthropic 96.8% 2026-11-11
15 GPT-6 Astra Non-reasoning OpenAI 96.7% 2026-11-11
16 GPT-6.1 Sol Extra High OpenAI 96.4% 2026-11-11
17 GPT-6.1 Sol Max OpenAI 96.2% 2026-11-11
18 GPT-6.1 Sol High OpenAI 95.0% 2026-11-11
19 Claude Fable 5.2 Max Anthropic 94.9% 2026-11-11
20 Claude Fable 5.2 Low Anthropic 92.3% 2026-11-11
21 GPT-6.1 Sol Medium OpenAI 91.0% 2026-11-11
22 GPT-6.1 Sol Low OpenAI 82.8% 2026-11-11
23 Claude Fable 5.2 Extra High Anthropic 81.2% 2026-11-11
24 Claude Fable 5.2 High Anthropic 79.3% 2026-11-11
25 Claude Fable 5.2 Medium Anthropic 73.3% 2026-11-11
26 Claude Sonnet 5.5 Extra High Anthropic 72.4% 2026-11-11
27 Claude Fable 5.2 Max Anthropic 65.7% 2026-11-11
28 Gemini 4 Carbon Low Google 65.1% 2026-11-11
29 Gemini 4 Carbon Medium Google 63.9% 2026-11-11
30 GPT-6 Astra Max OpenAI 62.7% 2026-11-11
31 GPT-6 Astra Extra High OpenAI 59.3% 2026-11-11
32 GPT-6 Astra High OpenAI 54.8% 2026-11-11
33 GPT-6.1 Sol Max OpenAI 52.7% 2026-11-11
34 Claude Fable 5.2 Low Anthropic 52.0% 2026-11-11
35 Gemini 4 Argon Low Google 49.8% 2026-11-11
36 Gemini 4 Argon Medium Google 46.3% 2026-11-11
37 GPT-6.1 Sol Extra High OpenAI 39.9% 2026-11-11
38 GPT-6 Astra Medium OpenAI 38.6% 2026-11-11
39 Claude Sonnet 5.5 High Anthropic 35.5% 2026-11-11
40 GPT-6 Astra Non-reasoning OpenAI 35.2% 2026-11-11
41 Gemini 3.8 Flash High Google 35.0% 2026-11-11
42 Claude Opus 5 High Anthropic 30.2% 2026-11-11
43 GPT-6.1 Sol High OpenAI 26.7% 2026-11-11
44 Claude Sonnet 5.5 Extra High Anthropic 24.3% 2026-11-11
45 Gemini 3.8 Flash Medium Google 24.2% 2026-11-11
46 GPT-6 Sol Max OpenAI 23.0% 2026-11-11
47 GPT-6 Astra Low OpenAI 17.5% 2026-11-11
48 Gemini 3.8 Flash Low Google 15.1% 2026-11-11
49 GPT-6.1 Sol Medium OpenAI 10.6% 2026-11-11
50 Gemini 3.8 Flash High Google 10.4% 2026-11-11
51 Grok 4.7 High xAI 10.1% 2026-11-11
52 GPT-6 Sol Extra High OpenAI 9.5% 2026-11-11
53 Grok 4.7 Medium xAI 8.4% 2026-11-11
54 GPT-5.6 Sol Max OpenAI 7.8% 2026-11-11
55 Grok 4.7 Extra High xAI 7.0% 2026-11-11
56 GPT-5.6 Sol Extra High OpenAI 7.0% 2026-11-11
57 Gemini 3.8 Flash Low Google 6.0% 2026-11-11
58 GPT-6 Sol High OpenAI 5.9% 2026-11-11
59 DeepSeek V4.1 Flash 1017 Max DeepSeek 5.8% 2026-11-11
60 GPT-6 Sol Medium OpenAI 4.9% 2026-11-11
61 Claude Sonnet 5.5 Medium Anthropic 4.9% 2026-11-11
62 GPT-6 Sol Max OpenAI 4.6% 2026-11-11
63 Gemini 3.8 Flash Medium Google 4.0% 2026-11-11
64 GPT-6.1 Sol Low OpenAI 3.9% 2026-11-11
65 DeepSeek V4.1 Flash Low DeepSeek 3.3% 2026-11-11
66 DeepSeek V4.1 Flash High DeepSeek 3.2% 2026-11-11
67 Claude Haiku 5.5 Max Anthropic 2.8% 2026-11-11
68 DeepSeek V4.1 Flash Max DeepSeek 2.7% 2026-11-11
69 GPT-5.6 Sol High OpenAI 2.2% 2026-11-11
70 Grok 4.6 Extra High xAI 2.1% 2026-11-11
71 Grok 4.7 Extra High xAI 1.8% 2026-11-11
72 GPT-6 Sol Extra High OpenAI 1.8% 2026-11-11
73 GPT-6 Sol Low OpenAI 1.8% 2026-11-11
74 Grok 4.7 High xAI 1.7% 2026-11-11
75 Grok 4.7 Medium xAI 1.7% 2026-11-11
76 Claude Opus 4.8 High Anthropic 1.5% 2026-11-11
77 GPT-6 Sol Non-reasoning OpenAI 1.2% 2026-11-11
78 Claude Sonnet 5.5 High Anthropic 1.2% 2026-11-11
79 GPT-5.6 Sol Medium OpenAI 1.1% 2026-11-11
80 GPT-6 Sol High OpenAI 0.9% 2026-11-11
81 DeepSeek V4.1 Flash Max DeepSeek 0.9% 2026-11-11
82 GPT-5.6 Terra Max OpenAI 0.8% 2026-11-11
83 Grok 4.7 Low xAI 0.8% 2026-11-11
84 GPT-5.6 Terra Extra High OpenAI 0.7% 2026-11-11
85 Claude Sonnet 5.5 Low Anthropic 0.6% 2026-11-11
86 GPT-6 Luna Max OpenAI 0.6% 2026-11-11
87 Claude Opus 4.6 Max Anthropic 0.5% 2026-11-11
88 GPT-6 Luna Extra High OpenAI 0.5% 2026-11-11
89 GPT-5.6 Terra High OpenAI 0.5% 2026-11-11
90 DeepSeek V4.1 Flash High DeepSeek 0.5% 2026-11-11
91 DeepSeek V4.1 Flash Non-reasoning DeepSeek 0.5% 2026-11-11
92 Claude Haiku 5.5 Max Anthropic 0.5% 2026-11-11
93 GPT-5.5 High OpenAI 0.4% 2026-11-11
94 Gemini 3.1 Pro Preview Google 0.4% 2026-11-11
95 Muse Spark Open Extra High Meta 0.4% 2026-11-11
96 GPT-6 Sol Medium OpenAI 0.4% 2026-11-11
97 Claude Sonnet 5.5 Medium Anthropic 0.4% 2026-11-11
98 Claude Sonnet 5.5 Low Anthropic 0.4% 2026-11-11
99 GPT-6 Luna High OpenAI 0.4% 2026-11-11
100 DeepSeek V4.1 Flash Low DeepSeek 0.4% 2026-11-11
101 DeepSeek V4.1 Flash Non-reasoning DeepSeek 0.3% 2026-11-11
102 GPT-5.6 Sol Low OpenAI 0.3% 2026-11-11
103 GPT-6 Luna Medium OpenAI 0.3% 2026-11-11
104 Grok 4.5 Medium xAI 0.3% 2026-11-11
105 Grok 4.5 High xAI 0.3% 2026-11-11
106 GPT-6 Sol Non-reasoning OpenAI 0.3% 2026-11-11
107 Grok 4.5 Low xAI 0.3% 2026-11-11
108 Grok 4.7 Low xAI 0.3% 2026-11-11
109 GPT-6 Luna Low OpenAI 0.2% 2026-11-11
110 DeepSeek V4.1 Flash 1017 Max DeepSeek 0.2% 2026-11-11
111 GPT-5.4 High OpenAI 0.2% 2026-11-11
112 GPT-6 Luna Medium OpenAI 0.2% 2026-11-11
113 GPT-5.6 Luna Max OpenAI 0.2% 2026-11-11
114 GPT-6 Luna High OpenAI 0.2% 2026-11-11
115 GPT-5.6 Luna Low OpenAI 0.2% 2026-11-11
116 GPT-5.6 Luna Medium OpenAI 0.2% 2026-11-11
117 GPT-6 Luna Extra High OpenAI 0.2% 2026-11-11
118 Muse Spark Open Extra High Meta 0.1% 2026-11-11
119 GPT-6 Sol Low OpenAI 0.1% 2026-11-11
120 GPT-6 Luna Max OpenAI 0.1% 2026-11-11
121 GPT-5.6 Luna High OpenAI 0.1% 2026-11-11
122 GPT-5.6 Terra Medium OpenAI 0.1% 2026-11-11
123 GPT-6 Luna Non-reasoning OpenAI 0.1% 2026-11-11
124 GPT-6 Luna Low OpenAI 0.0% 2026-11-11
125 GPT-6 Luna Non-reasoning OpenAI 0.0% 2026-11-11
126 GPT-5.6 Luna Extra High OpenAI 0.0% 2026-11-11
127 GPT-5.6 Terra Low OpenAI 0.0% 2026-11-11