ARC AI2

Models scored

134

Best score

0.95

Metric

Challenge score

ARC AI2 is an AI evaluation tracked by Epoch AI, with 134 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 November 2019 to December 2024. The best score recorded is 0.95, measured as Challenge score. Models come from 01.AI, Google Research, Technology Innovation Institute, OpenAI, Meta AI, MosaicML, Baichuan, Stability AI and others.

Full record

Score metric: Challenge score
Models scored: 134
Best score recorded: 0.95
Earliest model scored: Nov 5, 2019
Most recent model scored: Dec 26, 2024
Third-party reported: Yes

Leaderboard

01	Llama-3.1-405B	Meta AI	0.95
02	DeepSeek-V3	DeepSeek	0.95
03	Qwen2.5-72B	Alibaba	0.94
04	DeepSeek-V2	DeepSeek	0.92
05	Phi-3-medium-128k-instruct	Microsoft	0.92
06	Phi-3-small-8k-instruct	Microsoft	0.91
07	gpt-3.5-turbo-1106	OpenAI	0.87
08	Mixtral-8x7B-v0.1	Mistral AI	0.87
09	claude-instant-1.2	Anthropic	0.86
10	StableBeluga2	Stability AI	0.86
11	claude-instant-1.1	Anthropic	0.86
12	PaLM 540B	Google Research	0.85
13	text-davinci-002	OpenAI	0.85
14	Phi-3-mini-4k-instruct	Microsoft	0.85
15	Qwen-14B	Alibaba	0.84
16	Meta-Llama-3-8B-Instruct	Meta AI	0.83
17	internlm-20b		0.82
18	Mistral-7B-v0.1	Mistral AI	0.79
19	gemma-7b	Google DeepMind	0.78
20	Llama-2-70b-hf	Meta AI	0.78
21	phi-2	Microsoft	0.76
22	Qwen-7B	Alibaba	0.75
23	Qwen2.5-Coder-32B	Alibaba	0.7
24	LLaMA-65B	Meta AI	0.69
25	internlm-7b		0.69
26	PaLM 2-L		0.69
27	falcon-180B	Technology Innovation Institute	0.68
28	LLaMA-33B	Meta AI	0.68
29	Qwen2.5-Coder-14B		0.66
30	PaLM 2-M		0.65
31	DeepSeek-Coder-V2-Base	DeepSeek	0.64
32	falcon-40b	Technology Innovation Institute	0.62
33	chatglm2-6b		0.61
34	Qwen2.5-Coder-7B	Alibaba	0.61
35	Llama-2-13b	Meta AI	0.6
36	PaLM 2-S		0.6
37	DeepSeek-Coder-V2-Lite-Base		0.57
38	Yi-9B		0.56
39	Nemotron-4 15B	NVIDIA	0.56
40	INTELLECT-1-Instruct	Prime Intellect,Hugging Face,Arcee AI	0.55

More benchmarks

Epoch Capabilities Index MMLU Terminal-Bench GPQA Diamond OTIS Mock AIME 2024-2025 GSM8K ARC-AGI ARC-AGI-2

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