Llama 3.3 70B Instruct

Meta Llama

Llama 3.3 is a state-of-the-art multilingual large language model expertly designed for engaging in conversations across numerous languages. This powerful generative model boasts 70 billion parameters, allowing it to surpass the performance of many other publicly available and proprietary chat models on standard industry evaluations. With an extensive context window of 128,000 tokens, Llama 3.3 is well-suited for both commercial applications and research endeavors in a variety of languages.

Model Specifications

Technical details and capabilities of Llama 3.3 70B Instruct

Core Specifications

70.0B Parameters

Model size and complexity

15000.0B Training Tokens

Amount of data used in training

128.0K / 128.0K

Input / Output tokens

November 30, 2023

Knowledge cutoff date

December 5, 2024

Release date

Capabilities & License

Multimodal Support

Not Supported

Web Hydrated

No

License

Llama 3.3 Community License Agreement

Resources

API Reference

https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct

Playground

https://huggingface.co/meta-llama/Llama-3.3-70B-Instruct

Code Repository

https://github.com/meta-llama/llama-models/blob/main/models/llama3_3/MODEL_CARD.md

Performance Insights

Check out how Llama 3.3 70B Instruct handles various AI tasks through comprehensive benchmark results.

100

75

50

25

0

92.1

IFEval

92.1

(92%)

91.1

MGSM

91.1

(91%)

88.4

HumanEval

88.4

(88%)

87.6

MBPP EvalPlus

87.6

(88%)

86

MMLU

86

(86%)

77.3

BFCL v2

77.3

(77%)

77

MATH

77

(77%)

68.9

MMLU-Pro

68.9

(69%)

65.8

Arena Hard

65.8

(66%)

61.7

CRAG

61.7

(62%)

52.8

HELMET LongQA

52.8

(53%)

50.5

GPQA

50.5

(51%)

21.7

LongBench

21.7

(22%)

20

FinanceBench

20

(20%)

IFEval

MGSM

HumanEval

MBPP EvalPlus

MMLU

BFCL v2

MATH

MMLU-Pro

Arena Hard

CRAG

HELMET LongQA

GPQA

LongBench

FinanceBench

Model Comparison

See how Llama 3.3 70B Instruct stacks up against other leading models across key performance metrics.

100

80

60

40

20

0

86

MMLU - Llama 3.3 70B Instruct

86

(86%)

87.3

MMLU - Llama 3.1 405B Instruct

87.3

(87%)

86.2

MMLU - Grok-2 mini

86.2

(86%)

88.7

MMLU - GPT-4o

88.7

(89%)

83.3

MMLU - Qwen2.5 32B Instruct

83.3

(83%)

87.5

MMLU - Grok-2

87.5

(88%)

68.9

MMLU-Pro - Llama 3.3 70B Instruct

68.9

(69%)

73.3

MMLU-Pro - Llama 3.1 405B Instruct

73.3

(73%)

72

MMLU-Pro - Grok-2 mini

72

(72%)

72.6

MMLU-Pro - GPT-4o

72.6

(73%)

69

MMLU-Pro - Qwen2.5 32B Instruct

69

(69%)

75.5

MMLU-Pro - Grok-2

75.5

(76%)

50.5

GPQA - Llama 3.3 70B Instruct

50.5

(51%)

50.7

GPQA - Llama 3.1 405B Instruct

50.7

(51%)

51

GPQA - Grok-2 mini

51

(51%)

53.6

GPQA - GPT-4o

53.6

(54%)

49.5

GPQA - Qwen2.5 32B Instruct

49.5

(50%)

56.0

GPQA - Grok-2

56.0

(56%)

88.4

HumanEval - Llama 3.3 70B Instruct

88.4

(88%)

89

HumanEval - Llama 3.1 405B Instruct

89

(89%)

85.7

HumanEval - Grok-2 mini

85.7

(86%)

90.2

HumanEval - GPT-4o

90.2

(90%)

88.4

HumanEval - Qwen2.5 32B Instruct

88.4

(88%)

88.4

HumanEval - Grok-2

88.4

(88%)

77

MATH - Llama 3.3 70B Instruct

77

(77%)

73.8

MATH - Llama 3.1 405B Instruct

73.8

(74%)

73

MATH - Grok-2 mini

73

(73%)

76.6

MATH - GPT-4o

76.6

(77%)

83.1

MATH - Qwen2.5 32B Instruct

83.1

(83%)

76.1

MATH - Grok-2

76.1

(76%)

MMLU

MMLU-Pro

GPQA

HumanEval

MATH

Llama 3.3 70B Instruct

Llama 3.1 405B Instruct

Grok-2 mini

GPT-4o

Qwen2.5 32B Instruct

Grok-2

Detailed Benchmarks

Dive deeper into Llama 3.3 70B Instruct's performance across specific task categories. Expand each section to see detailed metrics and comparisons.

Coding

HumanEval

Claude 3.5 Sonnet

93.7%

89.0%

89.0%

Llama 3.1 405B Instruct

89.0%

Mistral Small 3.1 24B

88.4%

Llama 3.3 70B Instruct

88.4%

Qwen2.5 32B Instruct

88.4%

Qwen2.5-Coder 7B Instruct

88.4%

88.4%

Ministral 8B Instruct

34.8%

Current model

Other models

Avg (83.8%)

Knowledge

MMLU

91.8%

86.5%

86.4%

86.2%

Llama 3.2 90B Instruct

86.0%

Llama 3.3 70B Instruct

86.0%

85.9%

85.9%

85.2%

Llama 3.2 3B Instruct

63.4%

Current model

Other models

Avg (84.3%)

GPQA

87.7%

53.6%

Gemini 1.5 Flash

51.0%

51.0%

Llama 3.1 405B Instruct

50.7%

Llama 3.3 70B Instruct

50.5%

50.4%

Qwen2.5 32B Instruct

49.5%

Qwen2.5 72B Instruct

49.0%

Qwen2 7B Instruct

25.3%

Current model

Other models

Avg (51.9%)

MATH

97.9%

Qwen2.5 14B Instruct

80.0%

Claude 3.5 Sonnet

78.3%

78.0%

Gemini 1.5 Flash

77.9%

Llama 3.3 70B Instruct

77.0%

76.6%

76.6%

76.1%

32.6%

Current model

Other models

Avg (75.1%)

Non categorized

MMLU-Pro

84.0%

72.0%

Qwen2.5 72B Instruct

71.1%

70.4%

Qwen2.5 32B Instruct

69.0%

Llama 3.3 70B Instruct

68.9%

68.5%

Gemini 1.5 Flash

67.3%

Llama 3.1 70B Instruct

66.4%

Qwen2.5-Coder 7B Instruct

40.1%

Current model

Other models

Avg (67.8%)

IFEval

Claude 3.7 Sonnet

93.2%

92.1%

Llama 3.3 70B Instruct

92.1%

90.4%

90.0%

89.7%

Llama 3.1 405B Instruct

88.6%

Llama 3.1 70B Instruct

87.5%

61.3%

Current model

Other models

Avg (87.2%)

MBPP EvalPlus

Llama 3.1 405B Instruct

88.6%

Llama 3.3 70B Instruct

87.6%

Current model

Other models

Avg (88.1%)

MGSM

92.0%

Claude 3.5 Sonnet

91.6%

Claude 3.5 Sonnet

91.6%

Llama 3.3 70B Instruct

91.1%

90.8%

90.7%

90.5%

89.3%

Phi-3.5-mini-instruct

47.9%

Current model

Other models

Avg (86.2%)

Arena Hard

Mistral Small 3

87.6%

Jamba 1.6 Large

76.5%

Mistral Large 2

71.5%

Ministral 8B Instruct

70.9%

Llama 3.3 70B Instruct

65.8%

Jamba 1.5 Large

65.4%

51.2%

46.1%

Llama 3.1 8B Instruct

28.2%

Current model

Other models

Avg (62.6%)

CRAG

Jamba 1.6 Large

78.2%

76.2%

Mistral Large 2

67.0%

Llama 3.3 70B Instruct

61.7%

Llama 3.1 8B Instruct

60.0%

Ministral 8B Instruct

52.0%

50.3%

49.4%

43.1%

Current model

Other models

Avg (59.8%)

FinanceBench

Jamba 1.6 Large

64.5%

45.4%

Llama 3.1 8B Instruct

28.4%

Llama 3.3 70B Instruct

20.0%

Ministral 8B Instruct

19.2%

13.3%

Current model

Other models

Avg (31.8%)

HELMET LongQA

Jamba 1.6 Large

56.7%

Llama 3.3 70B Instruct

52.8%

46.9%

44.8%

Ministral 8B Instruct

33.0%

Llama 3.1 8B Instruct

29.2%

Mistral Large 2

26.7%

Current model

Other models

Avg (41.4%)

LongBench

Jamba 1.6 Large

38.8%

Mistral Large 2

34.4%

32.0%

Llama 3.3 70B Instruct

21.7%

18.9%

Llama 3.1 8B Instruct

17.7%

Ministral 8B Instruct

17.5%

Current model

Other models

Avg (25.9%)

Providers Pricing Coming Soon

We're working on gathering comprehensive pricing data from all major providers for Llama 3.3 70B 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

Stay Ahead with AI Updates

Get insights on Gemini Pro 2.5, Sonnet 3.7 and more top AI models