Codestral: Complete Benchmark Performance Guide

mistral codestral 2024-05-29 MNPL

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

Key Specifications

Vendormistral
Versioncodestral
Release Date2024-05-29
Context Window32000 tokens
Input Modalitiestext
Output Modalitiestext
LicenseMNPL
Documentationhttps://docs.mistral.ai/

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 75.1 % 2024-05-29 5-shot view
HUMANEVAL 69.2 pass@1 2024-05-29 — view
GSM8K 63.3 % 2024-05-29 0-shot CoT view
MATH 29.4 % 2024-05-29 0-shot CoT view
BBH 58.3 % 2024-05-29 3-shot CoT view
GPQA 32.4 % 2024-05-29 0-shot view
IFEVAL 59.4 % 2024-05-29 prompt_strict view
ARC 87.4 % 2024-05-29 challenge view
MUSR 47.2 % 2024-05-29 0-shot view
WINOGRANDE 75.8 % 2024-05-29 0-shot view

Pricing

Tier Price Currency
Input$0.3 / MtokUSD
Output$0.9 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://mistral.ai/technology/ · as of 2024-05-29

Compliance

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

Codestral

Model Overview

Mistral Codestral 22B 代码专用模型, 32K 上下文, 支持 80+ 编程语言, 代码生成与补全性能突出。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Mistral codestral 2024-05-29 32K text text MNPL

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 75.1 % 5-shot
HumanEval 69.2 pass@1 —
GSM8K (Grade School Math 8K) 63.3 % 0-shot CoT
MATH 29.4 % 0-shot CoT
BBH (BIG-Bench Hard) 58.3 % 3-shot CoT
GPQA 32.4 % 0-shot
IFEval 59.4 % prompt_strict
ARC 87.4 % challenge
MUSR 47.2 % 0-shot
WinoGrande 75.8 % 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