Qwen2 72B Instruct

Qwen

Qwen2-72B-Instruct is a 72 billion parameter language model fine-tuned for instruction following. Boasting an extensive context window of 131,072 tokens, this model is part of the cutting-edge Qwen2 series. Demonstrating exceptional performance, it outperforms the majority of open-source models and rivals the capabilities of proprietary models in numerous benchmark evaluations.

Model Specifications

Technical details and capabilities of Qwen2 72B Instruct

Core Specifications

72.0B Parameters

Model size and complexity

131.1K / 131.1K

Input / Output tokens

July 22, 2024

Release date

Capabilities & License

Multimodal Support

Not Supported

Web Hydrated

License

tongyi-qianwen

Resources

Research Paper

https://arxiv.org/abs/2309.00071

API Reference

https://huggingface.co/Qwen/Qwen2-72B

Playground

https://huggingface.co/Qwen/Qwen2-72B

Code Repository

https://huggingface.co/Qwen/Qwen2-72B

Performance Insights

Check out how Qwen2 72B Instruct handles various AI tasks through comprehensive benchmark results.

100

91.1

GSM8K

91.1

(91%)

90.1

CMMLU

90.1

(90%)

87.6

HellaSwag

87.6

(88%)

HumanEval

(86%)

85.1

Winogrande

85.1

(85%)

83.8

C-Eval

83.8

(84%)

82.4

BBH

82.4

(82%)

82.3

MMLU

82.3

(82%)

80.2

MBPP

80.2

(80%)

EvalPlus

(79%)

69.2

MultiPL-E

69.2

(69%)

68.9

ARC Challenge

68.9

(69%)

64.4

MMLU-Pro

64.4

(64%)

59.7

MATH

59.7

(60%)

54.8

TruthfulQA

54.8

(55%)

44.4

TheroemQA

44.4

(44%)

42.4

GPQA

42.4

(42%)

GSM8K

CMMLU

HellaSwag

HumanEval

Winogrande

C-Eval

BBH

MMLU

MBPP

EvalPlus

MultiPL-E

ARC Challenge

MMLU-Pro

MATH

TruthfulQA

TheroemQA

GPQA

Model Comparison

See how Qwen2 72B Instruct stacks up against other leading models across key performance metrics.

82.3

MMLU - Qwen2 72B Instruct

82.3

(91%)

86.8

MMLU - Claude 3 Opus

86.8

(96%)

86.2

MMLU - Grok-2 mini

86.2

(96%)

78.9

MMLU - Phi-3.5-MoE-instruct

78.9

(88%)

87.3

MMLU - Llama 3.1 405B Instruct

87.3

(97%)

MMLU - Llama 3.3 70B Instruct

(96%)

64.4

MMLU-Pro - Qwen2 72B Instruct

64.4

(72%)

68.5

MMLU-Pro - Claude 3 Opus

68.5

(76%)

MMLU-Pro - Grok-2 mini

(80%)

54.3

MMLU-Pro - Phi-3.5-MoE-instruct

54.3

(60%)

73.3

MMLU-Pro - Llama 3.1 405B Instruct

73.3

(81%)

68.9

MMLU-Pro - Llama 3.3 70B Instruct

68.9

(77%)

42.4

GPQA - Qwen2 72B Instruct

42.4

(47%)

50.4

GPQA - Claude 3 Opus

50.4

(56%)

GPQA - Grok-2 mini

(57%)

36.8

GPQA - Phi-3.5-MoE-instruct

36.8

(41%)

50.7

GPQA - Llama 3.1 405B Instruct

50.7

(56%)

50.5

GPQA - Llama 3.3 70B Instruct

50.5

(56%)

HumanEval - Qwen2 72B Instruct

(96%)

84.9

HumanEval - Claude 3 Opus

84.9

(94%)

85.7

HumanEval - Grok-2 mini

85.7

(95%)

70.7

HumanEval - Phi-3.5-MoE-instruct

70.7

(79%)

HumanEval - Llama 3.1 405B Instruct

(99%)

88.4

HumanEval - Llama 3.3 70B Instruct

88.4

(98%)

59.7

MATH - Qwen2 72B Instruct

59.7

(66%)

60.1

MATH - Claude 3 Opus

60.1

(67%)

MATH - Grok-2 mini

(81%)

59.5

MATH - Phi-3.5-MoE-instruct

59.5

(66%)

73.8

MATH - Llama 3.1 405B Instruct

73.8

(82%)

MATH - Llama 3.3 70B Instruct

(86%)

MMLU

MMLU-Pro

GPQA

HumanEval

MATH

Qwen2 72B Instruct

Claude 3 Opus

Grok-2 mini

Phi-3.5-MoE-instruct

Llama 3.1 405B Instruct

Llama 3.3 70B Instruct

Detailed Benchmarks

Dive deeper into Qwen2 72B Instruct's performance across specific task categories. Expand each section to see detailed metrics and comparisons.

Math

GSM8K

97.1%

Claude 3 Sonnet

92.3%

Qwen2.5 7B Instruct

91.6%

Llama 3.1 Nemotron 70B Instruct

91.4%

Qwen2.5-Coder 32B Instruct

91.1%

91.1%

90.8%

90.0%

88.9%

68.6%

Current model

Other models

Avg (89.3%)

Coding

HumanEval

93.7%

87.8%

87.2%

87.1%

86.6%

86.0%

85.7%

85.4%

84.9%

Ministral 8B Instruct

34.8%

Current model

Other models

Avg (81.9%)

MBPP

Qwen2.5-Coder 32B Instruct

90.2%

Qwen2.5 32B Instruct

84.0%

Qwen2.5-Coder 7B Instruct

83.5%

82.0%

80.8%

80.2%

79.2%

78.2%

74.4%

52.4%

Current model

Other models

Avg (78.5%)

Reasoning

HellaSwag

95.4%

95.3%

89.0%

88.6%

87.6%

86.4%

85.9%

85.2%

Phi-3.5-mini-instruct

69.4%

Current model

Other models

Avg (87.0%)

Knowledge

MMLU

91.8%

Mistral Large 2

84.0%

Command A

84.0%

Llama 3.1 70B Instruct

83.6%

83.3%

82.3%

82.0%

81.3%

81.2%

Llama 3.2 3B Instruct

63.4%

Current model

Other models

Avg (81.7%)

GPQA

87.7%

Nova Pro

46.9%

Llama 3.2 90B Instruct

46.7%

45.5%

45.3%

42.4%

42.0%

Llama 3.1 70B Instruct

41.7%

Claude 3.5 Haiku

41.6%

Qwen2 7B Instruct

25.3%

Current model

Other models

Avg (46.5%)

MATH

o3-mini

97.9%

Mistral Small 3.1 24B

69.3%

Llama 3.2 90B Instruct

68.0%

61.6%

60.1%

59.7%

59.5%

58.7%

Qwen2.5-Coder 32B Instruct

57.2%

Gemini 1.0 Pro

32.6%

Current model

Other models

Avg (62.5%)

Hallucination

TruthfulQA

Phi-3.5-MoE-instruct

77.5%

Llama 3.1 Nemotron 70B Instruct

58.6%

58.4%

58.3%

57.8%

54.8%

Qwen2.5-Coder 32B Instruct

54.2%

Jamba 1.5 Mini

54.1%

Qwen2.5-Coder 7B Instruct

50.6%

Mistral NeMo Instruct

50.3%

Current model

Other models

Avg (57.5%)

Non categorized

MMLU-Pro

DeepSeek-R1

84.0%

Gemini 1.5 Flash

67.3%

Llama 3.1 70B Instruct

66.4%

66.3%

65.0%

64.4%

63.7%

58.7%

56.8%

Qwen2.5-Coder 7B Instruct

40.1%

Current model

Other models

Avg (63.3%)

TheroemQA

Qwen2 72B Instruct

44.4%

Qwen2 7B Instruct

25.3%

Current model

Other models

Avg (34.9%)

BBH

87.6%

86.9%

84.5%

84.3%

82.4%

82.4%

79.5%

78.2%

70.6%

Current model

Other models

Avg (81.8%)

Winogrande

GPT-4

87.5%

Command R+

85.4%

Qwen2 72B Instruct

85.1%

Llama 3.1 Nemotron 70B Instruct

84.5%

Qwen2.5 32B Instruct

82.0%

Gemma 2 9B

80.6%

Mistral NeMo Instruct

76.8%

Ministral 8B Instruct

75.3%

Current model

Other models

Avg (82.2%)

ARC Challenge

Claude 3 Haiku

89.2%

Jamba 1.5 Mini

85.7%

Phi-3.5-mini-instruct

84.6%

Llama 3.1 8B Instruct

83.4%

Ministral 8B Instruct

71.9%

Command R+

71.0%

Llama 3.1 Nemotron 70B Instruct

69.2%

Qwen2 72B Instruct

68.9%

Current model

Other models

Avg (78.0%)

MultiPL-E

75.4%

75.1%

72.8%

70.4%

69.2%

59.1%

Current model

Other models

Avg (70.3%)

C-Eval

91.8%

86.5%

83.8%

77.2%

Current model

Other models

Avg (84.8%)

Providers Pricing Coming Soon

We're working on gathering comprehensive pricing data from all major providers for Qwen2 72B Instruct. Compare costs across platforms to find the best pricing for your use case.

OpenAI

Anthropic

Google

Mistral AI

Cohere

Share your feedback

Hi, I'm Charlie Palars, the founder of Deepranking.ai. I'm always looking for ways to improve the site and make it more useful for you. You can write me through this form or directly through X at @palarsio.

Your feedback helps us improve our service

Qwen2 72B Instruct

Model Specifications

Core Specifications

Capabilities & License

Resources

Performance Insights

Model Comparison

Detailed Benchmarks

Math

GSM8K

Coding

HumanEval

MBPP

Reasoning

HellaSwag

Knowledge

MMLU

GPQA

MATH

Hallucination

TruthfulQA

Non categorized

MMLU-Pro

TheroemQA

BBH

Winogrande

ARC Challenge

MultiPL-E

C-Eval

Providers Pricing Coming Soon

Share your feedback

Stay Ahead with AI Updates