Llama 3.1 70B: Complete Benchmark Performance Guide

meta 3.1-70b 2024-07-23 Llama 3 Community License

In-depth analysis of Llama 3.1 70B performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendormeta
Version3.1-70b
Release Date2024-07-23
Context Window128000 tokens
Input Modalitiestext
Output Modalitiestext
LicenseLlama 3 Community License
Documentationhttps://llama.meta.com/docs/

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 75.6 % 2024-07-23 5-shot view
HUMANEVAL 79.7 pass@1 2024-07-23 — view
GSM8K 78.8 % 2024-07-23 0-shot CoT view
MATH 38.5 % 2024-07-23 0-shot CoT view
BBH 70.2 % 2024-07-23 3-shot CoT view
GPQA 40 % 2024-07-23 0-shot view
IFEVAL 73.7 % 2024-07-23 prompt_strict view
ARC 92.3 % 2024-07-23 challenge view
MUSR 48.1 % 2024-07-23 0-shot view
WINOGRANDE 81 % 2024-07-23 0-shot view

Pricing

Tier Price Currency
Input$0.9 / MtokUSD
Output$0.9 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://ai.meta.com/blog/ · as of 2024-07-23

Compliance

  • Data Residency: self-host
  • SOC2: ✗
  • HIPAA: ✗
  • GDPR: ✗
  • ISO 27001: ✗

Llama 3.1 70B

Model Overview

Meta Llama 3.1 70B 中端开源模型, 128K 上下文, 性能接近 GPT-4, 推理与编码能力突出。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Meta 3.1-70b 2024-07-23 128K text text Llama 3 Community License

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 75.6 % 5-shot
HumanEval 79.7 pass@1 —
GSM8K (Grade School Math 8K) 78.8 % 0-shot CoT
MATH 38.5 % 0-shot CoT
BBH (BIG-Bench Hard) 70.2 % 3-shot CoT
GPQA 40.0 % 0-shot
IFEval 73.7 % prompt_strict
ARC 92.3 % challenge
MUSR 48.1 % 0-shot
WinoGrande 81.0 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • Reliable general-purpose model.

Weaknesses

  • Proprietary, not self-hostable.

Use Cases

  • Code generation and debugging

References