ARC-AGI-2

Models scored

159

Best score

0.93

Metric

Score

ARC-AGI-2 is an AI evaluation tracked by Epoch AI, with 159 scored model versions on record. The scores here are reported by third parties rather than produced by Epoch AI running the evaluation itself.

Scored models span July 2024 to July 2026. The best score recorded is 0.93, measured as Score. Models come from OpenAI, Anthropic, Google DeepMind, Google, xAI, Z.ai (Zhipu AI), Moonshot, MiniMax and others.

Full record

Score metric: Score
Models scored: 159
Best score recorded: 0.93
Earliest model scored: Jul 18, 2024
Most recent model scored: Jul 24, 2026
Third-party reported: Yes

Leaderboard

01	gpt-5.6-sol_max	OpenAI	0.93
02	gpt-5.6-sol_xhigh	OpenAI	0.9
03	claude-opus-5_max	Anthropic	0.88
04	gpt-5.6-sol_high	OpenAI	0.85
05	gpt-5.5_xhigh	OpenAI	0.85
06	gpt-5.5-pro_high	OpenAI	0.85
07	gemini-3-deep-think-preview		0.85
08	gpt-5.5-pro_xhigh	OpenAI	0.84
09	gpt-5.6-terra_max	OpenAI	0.84
10	gpt-5.4-pro-2026-03-05_xhigh	OpenAI	0.83
11	gpt-5.5_high	OpenAI	0.83
12	gemini-3.1-pro-preview	Google DeepMind	0.77
13	claude-opus-4-7_max	Anthropic	0.76
14	gpt-5.6-terra_xhigh	OpenAI	0.74
15	gpt-5.4-2026-03-05_xhigh	OpenAI	0.74
16	gemini-3.5-flash_high	Google	0.72
17	claude-opus-4-8_high	Anthropic	0.72
18	claude-opus-4-8_medium	Anthropic	0.72
19	gpt-5.5_medium	OpenAI	0.7
20	claude-opus-4-6_120K	Anthropic	0.69
21	claude-opus-4-7_high	Anthropic	0.68
22	claude-opus-4-7_medium	Anthropic	0.68
23	gpt-5.4-2026-03-05_high	OpenAI	0.68
24	gpt-5.6-terra_high	OpenAI	0.67
25	gpt-5.6-sol_medium	OpenAI	0.67
26	grok-4-20	xAI	0.65
27	claude-opus-4-8_low	Anthropic	0.62
28	claude-opus-4-7_low	Anthropic	0.62
29	claude-sonnet-4-6_high	Anthropic	0.6
30	gpt-5.6-luna_max	OpenAI	0.6
31	claude-sonnet-4-6_max	Anthropic	0.58
32	gpt-5.4-2026-03-05_medium	OpenAI	0.55
33	gpt-5.2-pro-2025-12-11_high	OpenAI	0.54
34	gpt-5.2-2025-12-11_xhigh	OpenAI	0.53
35	gpt-5.6-luna_xhigh	OpenAI	0.48
36	gpt-5.2-2025-12-11_high	OpenAI	0.43
37	gpt-5.6-sol_low	OpenAI	0.42
38	gpt-5.2-pro-2025-12-11_medium	OpenAI	0.38
39	claude-opus-4-5-20251101_64K	Anthropic	0.38
40	gpt-5.6-terra_medium	OpenAI	0.38
41	gemini-3-flash-preview	Google DeepMind	0.34
42	gpt-5.5_low	OpenAI	0.33
43	gemini-3-pro-preview	Google DeepMind	0.31
44	gpt-5.6-luna_high	OpenAI	0.29
45	gpt-5.4-2026-03-05_low	OpenAI	0.29
46	gpt-5.2-2025-12-11_medium	OpenAI	0.27
47	claude-opus-4-5-20251101_16K	Anthropic	0.23
48	glm-5.2_unknown	Z.ai (Zhipu AI)	0.23
49	gpt-5.4-mini-2026-03-17_xhigh	OpenAI	0.19
50	gpt-5.6-terra_low	OpenAI	0.19
51	gpt-5-pro-2025-10-06_unknown	OpenAI	0.18
52	gpt-5.1-2025-11-13_high	OpenAI	0.18
53	grok-4-0709	xAI	0.16
54	claude-opus-4-5-20251101_8K	Anthropic	0.14
55	claude-sonnet-4-5-20250929_32K	Anthropic	0.14

More benchmarks

Epoch Capabilities Index MMLU Terminal-Bench GPQA Diamond OTIS Mock AIME 2024-2025 GSM8K ARC-AGI WeirdML

SourceEpoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/benchmarks. Licensed under CC BY 4.0.

← All benchmarks