Mathstral 7B: Complete Benchmark Performance Guide

mistral mathstral-7b 2024-07-16 Apache 2.0

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

Key Specifications

Vendormistral
Versionmathstral-7b
Release Date2024-07-16
Context Window32768 tokens
Input Modalitiestext
Output Modalitiestext
LicenseApache 2.0
Documentationhttps://docs.mistral.ai/

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 69.1 % 2024-07-16 5-shot view
HUMANEVAL 49.9 pass@1 2024-07-16 — view
GSM8K 72.4 % 2024-07-16 0-shot CoT view
MATH 61 % 2024-07-16 0-shot CoT view
BBH 69.8 % 2024-07-16 3-shot CoT view
GPQA 37.5 % 2024-07-16 0-shot view
IFEVAL 63.6 % 2024-07-16 prompt_strict view
ARC 91.5 % 2024-07-16 challenge view
MUSR 42.3 % 2024-07-16 0-shot view
WINOGRANDE 71 % 2024-07-16 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 2024-07-16

Compliance

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

Mathstral 7B

Model Overview

Mistral Mathstral 7B 数学专用模型, 32K 上下文, 基于 Mistral 7B 微调, 在 MATH 基准上达到 56.6%。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Mistral mathstral-7b 2024-07-16 32K text text Apache 2.0

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 69.1 % 5-shot
HumanEval 49.9 pass@1 —
GSM8K (Grade School Math 8K) 72.4 % 0-shot CoT
MATH 61.0 % 0-shot CoT
BBH (BIG-Bench Hard) 69.8 % 3-shot CoT
GPQA 37.5 % 0-shot
IFEval 63.6 % prompt_strict
ARC 91.5 % challenge
MUSR 42.3 % 0-shot
WinoGrande 71.0 % 0-shot

Pricing

Input Output Cache Read Cache Write
— — — —

per million tokens

Strengths

  • Reliable general-purpose model.

Weaknesses

  • HumanEval 49.9, coding weak.
  • Proprietary, not self-hostable.

Use Cases

  • General chat and Q&A

References