GSM8K

Models scored

170

Best score

0.94

Metric

EM

GSM8K is an AI evaluation tracked by Epoch AI, with 170 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 June 2020 to November 2024. The best score recorded is 0.94, measured as EM. Models come from OpenAI, Meta AI, Mistral AI, MosaicML, Technology Innovation Institute, Large Model Systems Organization,University of California (UC) Berkeley, Hugging Face,BigScience, Google DeepMind and others.

Full record

Score metric: EM
Models scored: 170
Best score recorded: 0.94
Earliest model scored: Jun 22, 2020
Most recent model scored: Nov 29, 2024
Third-party reported: Yes

Leaderboard

01	DeepSeek-Coder-V2-Instruct	DeepSeek	0.94
02	Qwen2.5-Coder-14B-Instruct		0.94
03	Qwen2.5-Coder-32B-Instruct	Alibaba	0.93
04	gpt-4-0314	OpenAI	0.92
05	gpt-4o-mini-2024-07-18	OpenAI	0.91
06	Qwen2.5-Coder-32B	Alibaba	0.91
07	gpt-4-0613	OpenAI	0.9
08	Qwen2.5-Coder-14B		0.89
09	Phi-3.5-MoE-instruct	Microsoft	0.89
10	DeepSeek-Coder-V2-Lite-Instruct		0.88
11	claude-instant-1.2	Anthropic	0.87
12	Qwen2.5-Coder-7B-Instruct	Alibaba	0.87
13	Phi-3.5-mini-instruct	Microsoft	0.86
14	DeepSeek-Coder-V2-Base	DeepSeek	0.86
15	gemma-2-9b	Google DeepMind	0.85
16	Mistral-Nemo-Base-2407	Mistral AI	0.84
17	Qwen2.5-Coder-7B	Alibaba	0.84
18	Llama-3.1-8B-Instruct	Meta AI	0.82
19	gemini-1.5-flash-001	Google DeepMind	0.82
20	claude-instant-1.1	Anthropic	0.81
21	Qwen2.5-Coder-3B-Instruct		0.81
22	text-davinci-003		0.78
23	Yi-34B-Chat	01.AI	0.76
24	Qwen2.5-Coder-3B		0.76
25	Mixtral-8x7B-v0.1	Mistral AI	0.74
26	StableBeluga2	Stability AI	0.7
27	Llama-2-70b-hf	Meta AI	0.7
28	Yi-34B	01.AI	0.67
29	DeepSeek-Coder-V2-Lite-Base		0.67
30	Qwen2.5-Coder-1.5B	Alibaba	0.66
31	internlm-20b		0.63
32	Qwen-14B	Alibaba	0.61
33	Qwen-14B-Chat	Alibaba	0.61
34	Llama-2-70b-chat	Meta AI	0.59
35	gpt-3.5-turbo-0613	OpenAI	0.58
36	starcoder2-15b	Hugging Face,ServiceNow,NVIDIA,BigCode	0.58
37	code-davinci-002	OpenAI	0.57
38	PaLM 540B	Google Research	0.56
39	LLaMA-65B	Meta AI	0.54
40	Mistral-7B-v0.1	Mistral AI	0.54
41	falcon-180B	Technology Innovation Institute	0.54
42	falcon-11b	Technology Innovation Institute	0.54
43	Baichuan-2-13B-Base	Baichuan	0.53
44	Qwen-7B	Alibaba	0.52

More benchmarks

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