Llama 3.1 70B

#29Open Source

Benchmarks

Gray bar = dataset average

MMLU57 academic subjects

83.6%avg 81.8

GPQA DiamondPhD-level science questions

46.7%avg 50.4

HumanEvalPython code generation

80.5%avg 79.9

SWE-Bench VerifiedReal GitHub engineering tasks

21.8%avg 32.4

Composite ScoreScore = MMLU×20% + GPQA×30% + HumanEval×25% + SWE-Bench×25%56.3

HF Open LLM Leaderboard

IFEval · BBH · MATH · GPQA · MuSR · MMLU-Pro

Average

43.4

View on HF →

IFEvalInstruction following

86.7%avg 79.9

BBHBig Bench Hard

55.9%avg 48.7

MATH Lvl 5Competition mathematics

38.1%avg 37.8

GPQAGraduate-level science Q&A

14.2%avg 12.9

MuSRMultistep soft reasoning

17.7%avg 10.7

MMLU-ProProfessional knowledge

47.9%avg 40.0

Made by

🏢

Meta AI

Building AI for everyone

lab

→

More from Meta

Llama 3 8B InstructOSS

$0.140/1M→

Llama 3.1 8B InstructOSS

$0.020/1M→

Llama 3.1 70B InstructOSS

$0.400/1M→

Llama 4 Scout

$0.100/1M→

API Providers

Details

Provider: Meta
Parameters: 70B
Context: 128K tokens
Speed: 150 tok/s
Open Source: Yes
License: Open (varies)
Released: -

Reviews

No reviews yet

← View all 376 models