Mistral Small 3: Complete Benchmark Performance Guide

mistral small-3 2025-01-29 Apache 2.0

In-depth analysis of Mistral Small 3 performance on MMLU, HumanEval, GSM8K, MATH, BBH benchmarks, including pricing and use cases.

Key Specifications

Vendormistral
Versionsmall-3
Release Date2025-01-29
Context Window32768 tokens
Input Modalitiestext
Output Modalitiestext
LicenseApache 2.0
Documentationhttps://docs.mistral.ai/

Benchmark Performance

Benchmark Score Unit Evaluated At Notes Source
MMLU 76.9 % 2025-01-29 5-shot view
HUMANEVAL 74.4 pass@1 2025-01-29 — view
GSM8K 79.6 % 2025-01-29 0-shot CoT view
MATH 39.2 % 2025-01-29 0-shot CoT view
BBH 71.3 % 2025-01-29 3-shot CoT view
GPQA 30.1 % 2025-01-29 0-shot view
IFEVAL 76.9 % 2025-01-29 prompt_strict view
ARC 91.8 % 2025-01-29 challenge view
MUSR 46 % 2025-01-29 0-shot view
WINOGRANDE 78.8 % 2025-01-29 0-shot view

Pricing

Tier Price Currency
Input$0.2 / MtokUSD
Output$0.6 / MtokUSD
Cache Read$0 / MtokUSD
Cache Write$0 / MtokUSD

Source: https://mistral.ai/technology/ · as of 2025-01-29

Compliance

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

Mistral Small 3

Model Overview

Mistral Small 3 开源模型, 24B 参数, 32K 上下文, 性能接近 Mistral Large 2, Apache 2.0 可商用。

Core Specifications

Vendor Version Release Date Context Window Input Modalities Output Modalities License
Mistral small-3 2025-01-29 32K text text Apache 2.0

Benchmark Performance

Benchmark Score Unit Notes
MMLU (Massive Multitask Language Understanding) 76.9 % 5-shot
HumanEval 74.4 pass@1 —
GSM8K (Grade School Math 8K) 79.6 % 0-shot CoT
MATH 39.2 % 0-shot CoT
BBH (BIG-Bench Hard) 71.3 % 3-shot CoT
GPQA 30.1 % 0-shot
IFEval 76.9 % prompt_strict
ARC 91.8 % challenge
MUSR 46.0 % 0-shot
WinoGrande 78.8 % 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