Mistral Small 3.2

Overview

Freshness note: Model capabilities, limits, and pricing can change quickly. This profile is a point-in-time snapshot last verified on June 20, 2026.

Mistral Small 3.2 is Mistral AI’s June 2025 24B-class small model for production-grade instruction following and coding support at low operating cost. It is no longer the current Small-family default: Mistral’s current model overview marks it as deprecated as of April 30, 2026 and points users toward Mistral Small 4.

This page is retained for teams with existing mistral-small-2506 integrations, older benchmark notes, or migration work. New evaluations should generally start with Mistral Small 4 or Mistral Medium 3.5, depending on the required capability tier.

Capabilities

Mistral Small 3.2 was strongest for everyday assistant workloads: coding help, structured content generation, extraction/classification, and tool-augmented workflow orchestration. It remains relevant where an application is pinned to the older endpoint and needs predictable behavior during migration.

For new multilingual, coding, and low-cost assistant builds, Mistral Small 4 is the cleaner replacement because it is the current open Small-family route and has a larger published context window.

Technical Details

Official Mistral docs list a 128K context window for Small 3.2 and identify the API model as mistral-small-2506. The current docs also list a deprecation date of April 30, 2026 and Mistral Small 4 as the replacement.

Compared with many larger frontier models, Small 3.2 usually requires more retrieval/prompt structure for difficult reasoning tasks. That tradeoff made sense in cost-sensitive systems, but there is less reason to adopt it fresh now that Mistral Small 4 occupies the same practical lane.

Pricing & Access

This repo keeps the last verified pricing values for historical comparability:

• Input: $0.10 per 1M tokens
• Output: $0.30 per 1M tokens

Check Mistral’s current pricing and deprecation tables before running new traffic through the old endpoint, because availability can differ between API aliases, cloud partners, and self-managed weights.

Best Use Cases

Use Mistral Small 3.2 only when you are maintaining an existing pinned integration, comparing older benchmark runs, or staging a migration.

For fresh builds, use Mistral Small 4 when cost and latency dominate, or Mistral Medium 3.5 when the work needs stronger agentic coding and multimodal behavior.

Comparisons

• Mistral Small 4 (Mistral AI): Current replacement and better default for new Small-family evaluations.
• Mistral Medium 3.5 (Mistral AI): More capable current route for agentic coding, multimodal input, and harder reasoning.
• Qwen3.5 (Alibaba): Stronger Chinese-first bilingual alternative; Small 3.2 is now mainly a historical Mistral reference.