Which model leads on ARC-AGI-2?

GPT-6 Astra from OpenAI leads ARC-AGI-2 with a score of 95.0. The median score across 67 tested models is 16.0.

Is ARC-AGI-2 saturated?

No · the top score is 95.0 out of 100 (95%). There is still meaningful room for improvement on ARC-AGI-2.

Does ARC-AGI-2 predict performance on other benchmarks?

Yes · ARC-AGI-2 scores correlate 0.96 with Artificial Analysis · GPQA Diamond across 7 shared models. Models that do well on ARC-AGI-2 tend to do well on Artificial Analysis · GPQA Diamond.

How often is ARC-AGI-2 data refreshed?

BenchGecko pulls updates daily. New model scores on ARC-AGI-2 appear as soon as they are published by Epoch AI or the model provider.

Benchmark · ReasoningSettled

ARC-AGI-2

Name: ARC-AGI-2 Benchmark
Creator: BenchGecko
License: https://creativecommons.org/licenses/by/4.0/

ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.

Updated 2026-09-22

Models tested

Top score

95.0

GPT-6 Astra

Median

16.0

min 0.1

Top-5 spread

σ 1.8

Settled

The Frontier

Best score over time · one chart, every benchmark

Chart type

Frontier on ARC-AGI-2 rose from 1.3 to 95.0 in 20 months · +93.7 points · latest leader GPT-6 Astra from OpenAI.

Pink dots = frontier records · 16 totalClick to open model page

Full rankings

67 models tested · sorted by score

#	Model	Score	Price
1	GPT-6 Astra· OpenAI	95.0	$10.00
2	Claude Opus 5.5· Anthropic	92.5	$4.00
3	GPT-5.6 Sol· OpenAI	92.5	$2.00
4	Claude Opus 5· Anthropic	90.4	$5.00
5	Claude Fable 5.1· Anthropic	90.0	$10.00
6	Claude Fable 5· Anthropic	89.2	$10.00
7	Gemini 3.7 Flash· Google DeepMind	84.6	$0.75
8	GPT-5.6 Terra· OpenAI	83.9	$2.00
9	GPT-5.4 Pro· OpenAI	83.3	$30.00
10	Gemini 3.1 Pro Preview· Google DeepMind	77.1	$2.00
11	Claude Opus 4.7· Anthropic	75.8	$5.00
12	GPT-5.4· OpenAI	74.0	$2.50
13	Claude Opus 4.8· Anthropic	72.1	$5.00
14	Gemini 3.5 Flash· Google DeepMind	72.1	$1.50
15	Claude Opus 4.6· Anthropic	69.2	$5.00
16	Grok 4.6· xAI	67.1	$2.00
17	Grok 4.20· xAI	65.1	$1.25
18	DeepSeek V4 Flash 0731· DeepSeek	61.4	$0.02
19	DeepSeek V4 Pro 0813· DeepSeek	61.3	$0.55
20	Claude Sonnet 4.6· Anthropic	60.4	$3.00
21	Gemini 3.6 Flash· Google DeepMind	60.4	$0.75
22	Kimi K3· moonshotai	60.4	$0.67
23	GPT-5.6 Luna· OpenAI	59.5	$0.20
24	GPT-5.2 Pro· OpenAI	54.2	$21.00
25	GPT-5.2· OpenAI	52.9	$1.75
26	Grok 4.5· xAI	52.6	$2.00
27	Claude Opus 4.5· Anthropic	37.6	$5.00
28	Gemini 3 Flash Preview· Google DeepMind	33.6	$0.50
29	Gemini 3 Pro· Google DeepMind	31.1	n/a
30	GLM 5.2· z-ai	22.8	$1.00
31	GPT-5.4 Mini· OpenAI	18.9	$0.75
32	GPT-5 Pro· OpenAI	18.3	$15.00
33	GPT-5.1· OpenAI	17.6	$1.25
34	Grok 4· xAI	16.0	n/a
35	Claude Sonnet 4.5· Anthropic	13.6	$3.00
36	Kimi K2.5· moonshotai	11.8	$0.45
37	Gemini 3.5 Flash Lite· Google DeepMind	10.3	$0.30
38	GPT-5· OpenAI	9.9	$1.25
39	Claude Opus 4· Anthropic	8.6	$15.00
40	o3· OpenAI	6.5	$2.00
41	o4 Mini· OpenAI	6.1	$1.10
42	Claude Sonnet 4· Anthropic	5.9	$3.00
43	GPT-5.4 Nano· OpenAI	5.7	$0.20
44	Grok 4 Fast· xAI	5.3	n/a
45	Gemini 2.5 Pro· Google DeepMind	4.9	$1.25
46	GLM 5· z-ai	4.9	$0.60
47	MiniMax M2.5· minimax	4.9	$0.27
48	o3 Pro· OpenAI	4.9	$20.00
49	GPT-5 Mini· OpenAI	4.4	$0.25
50	Claude Haiku 4.5· Anthropic	4.0	$1.00
51	DeepSeek V3.2· DeepSeek	4.0	$0.28
52	o3 Mini· OpenAI	3.0	$1.10
53	GPT-5 Nano· OpenAI	2.6	$0.05
54	Gemini 2.5 Flash· Google DeepMind	2.5	$0.30
55	Gemini 2.0 Flash· Google DeepMind	1.3	$0.10
56	Gemini 2.0 Flash (Dec 2024)· Google DeepMind	1.3	n/a
57	R1· DeepSeek	1.3	$0.70
58	Qwen3 235B A22B Instruct 2507· Alibaba Qwen	1.3	$0.09
59	R1 0528· DeepSeek	1.1	$0.50
60	Claude 3.7 Sonnet· Anthropic	0.9	n/a
61	o1-mini· OpenAI	0.8	n/a
62	Gemini 1.5 Pro (Feb 2024)· Google DeepMind	0.8	n/a
63	Gemini 1.5 Pro (May 2024)· Google DeepMind	0.8	n/a
64	GPT-4.5· OpenAI	0.8	n/a
65	GPT-4.1· OpenAI	0.4	$2.00
66	Grok-3 mini· xAI	0.4	n/a
67	GPT-4o-mini (2024-07-18)· OpenAI	0.1	$0.15