Mistral 7B v0.2: Complete Benchmark Performance Guide

mistral 7b-v0.2 2024-03-23 Apache 2.0

In-depth analysis of Mistral 7B v0.2 performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendormistral
Version7b-v0.2
Release Date2024-03-23
Context Window32768 tokens
Input Modalitiestext
Output Modalitiestext
LicenseApache 2.0
Documentationhttps://docs.mistral.ai/

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 64.4 % 2024-03-23 5-shot view
HUMANEVAL 59.7 pass@1 2024-03-23 — view
GSM8K 56.2 % 2024-03-23 0-shot CoT view
MATH 25.9 % 2024-03-23 0-shot CoT view
BBH 71.3 % 2024-03-23 3-shot CoT view
GPQA 33.9 % 2024-03-23 0-shot view
IFEVAL 59.2 % 2024-03-23 prompt_strict view
ARC 85.5 % 2024-03-23 challenge view
MUSR 36.7 % 2024-03-23 0-shot view
WINOGRANDE 71.7 % 2024-03-23 0-shot view

Pricing

Tier Price Currency
Input$0.18 / MtokUSD
Output$0.18 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://mistral.ai/technology/ · as of 2024-03-23

Compliance

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

Mistral 7B v0.2

Model Overview

Mistral 7B v0.2 开源模型, 32K 上下文, 扩展上下文窗口, 改进推理, 适合通用任务。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Mistral 7b-v0.2 2024-03-23 32K text text Apache 2.0

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 64.4 % 5-shot
HumanEval 59.7 pass@1 —
GSM8K (Grade School Math 8K) 56.2 % 0-shot CoT
MATH 25.9 % 0-shot CoT
BBH (BIG-Bench Hard) 71.3 % 3-shot CoT
GPQA 33.9 % 0-shot
IFEval 59.2 % prompt_strict
ARC 85.5 % challenge
MUSR 36.7 % 0-shot
WinoGrande 71.7 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • Reliable general-purpose model.

Weaknesses

  • Proprietary, not self-hostable.

Use Cases

  • General chat and Q&A

References