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

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU75.1%2024-05-295-shotview
HUMANEVAL69.2pass@12024-05-29—view
GSM8K63.3%2024-05-290-shot CoTview
MATH29.4%2024-05-290-shot CoTview
BBH58.3%2024-05-293-shot CoTview
GPQA32.4%2024-05-290-shotview
IFEVAL59.4%2024-05-29prompt_strictview
ARC87.4%2024-05-29challengeview
MUSR47.2%2024-05-290-shotview
WINOGRANDE75.8%2024-05-290-shotview

Pricing

TierPriceCurrency
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

모델 개요

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。

핵심 사양

공급업체버전출시일컨텍스트 창입력 모달리티출력 모달리티라이선스
Mistralcodestral2024-05-2932KtexttextMNPL

벤치마크 성능

벤치마크점수단위비고
MMLU (Massive Multitask Language Understanding)75.1%5-shot
HumanEval69.2pass@1—
GSM8K (Grade School Math 8K)63.3%0-shot CoT
MATH29.4%0-shot CoT
BBH (BIG-Bench Hard)58.3%3-shot CoT
GPQA32.4%0-shot
IFEval59.4%prompt_strict
ARC87.4%challenge
MUSR47.2%0-shot
WinoGrande75.8%0-shot

가격

입력출력캐시 읽기캐시 쓰기
————

백만 토큰당

강점

  • 可靠的通用模型。

약점

  • 闭源专有模型，不支持自托管。

사용 사례

  • 代码生成与调试

참고문헌