Mixtral 8x7B: Complete Benchmark Performance Guide

mistral 8x7b 2023-12-11 Apache 2.0

In-depth analysis of Mixtral 8x7B performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendormistral
Version8x7b
Release Date2023-12-11
Context Window32000 tokens
Input Modalitiestext
Output Modalitiestext
LicenseApache 2.0
Documentationhttps://docs.mistral.ai/

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 77.2 % 2023-12-11 5-shot view
HUMANEVAL 79 pass@1 2023-12-11 — view
GSM8K 79.7 % 2023-12-11 0-shot CoT view
MATH 38 % 2023-12-11 0-shot CoT view
BBH 78.4 % 2023-12-11 3-shot CoT view
GPQA 43 % 2023-12-11 0-shot view
IFEVAL 72.1 % 2023-12-11 prompt_strict view
ARC 91.6 % 2023-12-11 challenge view
MUSR 53.5 % 2023-12-11 0-shot view
WINOGRANDE 82 % 2023-12-11 0-shot view

Pricing

Tier Price Currency
Input$0.7 / MtokUSD
Output$0.7 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://mistral.ai/technology/ · as of 2023-12-11

Compliance

  • Data Residency: EU
  • SOC2: ✓
  • HIPAA: ✗
  • GDPR: ✓
  • ISO 27001: ✓

Mixtral 8x7B

Model Overview

Mistral Mixtral 8x7B 首个开源 MoE 模型, 总参 47B/活跃 13B, 32K 上下文, 性能接近 GPT-3.5。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Mistral 8x7b 2023-12-11 32K text text Apache 2.0

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 77.2 % 5-shot
HumanEval 79.0 pass@1 —
GSM8K (Grade School Math 8K) 79.7 % 0-shot CoT
MATH 38.0 % 0-shot CoT
BBH (BIG-Bench Hard) 78.4 % 3-shot CoT
GPQA 43.0 % 0-shot
IFEval 72.1 % prompt_strict
ARC 91.6 % challenge
MUSR 53.5 % 0-shot
WinoGrande 82.0 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • Mixture-of-Experts architecture.

Weaknesses

  • Proprietary, not self-hostable.

Use Cases

  • Code generation and debugging

References