Mistral Large: Complete Benchmark Performance Guide

mistral large 2024-02-26 Apache 2.0

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

Key Specifications

Vendormistral
Versionlarge
Release Date2024-02-26
Context Window32000 tokens
Input Modalitiestext
Output Modalitiestext
LicenseApache 2.0
Documentationhttps://docs.mistral.ai/

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 82 % 2024-02-26 5-shot view
HUMANEVAL 73.4 pass@1 2024-02-26 — view
GSM8K 80.9 % 2024-02-26 0-shot CoT view
MATH 49.3 % 2024-02-26 0-shot CoT view
BBH 76.7 % 2024-02-26 3-shot CoT view
GPQA 38.1 % 2024-02-26 0-shot view
IFEVAL 75.8 % 2024-02-26 prompt_strict view
ARC 93.3 % 2024-02-26 challenge view
MUSR 52.1 % 2024-02-26 0-shot view
WINOGRANDE 83.3 % 2024-02-26 0-shot view

Pricing

Tier Price Currency
Input$4 / MtokUSD
Output$12 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://mistral.ai/technology/ · as of 2024-02-26

Compliance

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

Mistral Large

Model Overview

Mistral AI Large 首代旗舰, 32K 上下文, 多语言与推理能力突出, 原生支持函数调用与 JSON 输出。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Mistral large 2024-02-26 32K text text Apache 2.0

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 82.0 % 5-shot
HumanEval 73.4 pass@1 —
GSM8K (Grade School Math 8K) 80.9 % 0-shot CoT
MATH 49.3 % 0-shot CoT
BBH (BIG-Bench Hard) 76.7 % 3-shot CoT
GPQA 38.1 % 0-shot
IFEval 75.8 % prompt_strict
ARC 93.3 % challenge
MUSR 52.1 % 0-shot
WinoGrande 83.3 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • MMLU score 82.0, strong knowledge reasoning.

Weaknesses

  • Proprietary, not self-hostable.

Use Cases

  • Code generation and debugging
  • Agent workflows and tool use

References