Key Specifications

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

Benchmark Performance

BenchmarkScoreUnitEvaluated AtNotesSource
MMLU77.2%2023-12-115-shotview
HUMANEVAL79pass@12023-12-11view
GSM8K79.7%2023-12-110-shot CoTview
MATH38%2023-12-110-shot CoTview
BBH78.4%2023-12-113-shot CoTview
GPQA43%2023-12-110-shotview
IFEVAL72.1%2023-12-11prompt_strictview
ARC91.6%2023-12-11challengeview
MUSR53.5%2023-12-110-shotview
WINOGRANDE82%2023-12-110-shotview

Pricing

TierPriceCurrency
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

모델 개요

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。

핵심 사양

공급업체버전출시일컨텍스트 창입력 모달리티출력 모달리티라이선스
Mistral8x7b2023-12-1132KtexttextApache 2.0

벤치마크 성능

벤치마크점수단위비고
MMLU (Massive Multitask Language Understanding)77.2%5-shot
HumanEval79.0pass@1
GSM8K (Grade School Math 8K)79.7%0-shot CoT
MATH38.0%0-shot CoT
BBH (BIG-Bench Hard)78.4%3-shot CoT
GPQA43.0%0-shot
IFEval72.1%prompt_strict
ARC91.6%challenge
MUSR53.5%0-shot
WinoGrande82.0%0-shot

가격

입력출력캐시 읽기캐시 쓰기

백만 토큰당

강점

  • 采用 MoE 混合专家架构。

약점

  • 闭源专有模型,不支持自托管。

사용 사례

  • 代码生成与调试

참고문헌