Scan to contactMistral Small
Mistral Small for low-latency high-concurrency and cost-sensitive workloads
Mistral Small is a Mistral model for low-latency high-concurrency and cost-sensitive workloads, with options across enterprise, coding, vision and open-weight workflows.
starsCapabilities
visibilityVision understandingcodeFunction callingstreamStreaming output
paymentsContext and pricing
Context limit256,000
Max output256,000
Knowledge cutoff2025-06
Input price$0.15/ 1M tokens
Output price$0.6/ 1M tokens
descriptionOverview
Overview
Mistral Small is listed in Mistral's official model catalog, with model ID mistral-small-latest. Mistral's lineup spans advanced general models, lightweight models, coding models, vision models and open-weight MoE models.
Best for
Use Mistral Small when your product needs multilingual capability, European vendor options, coding assistance, vision understanding or controlled deployment. Test multilingual quality, reasoning, deployment method, cost and compliance requirements before production.
lightbulbUse cases
- Multilingual assistants and enterprise Q&A
- Code completion and developer help
- Multimodal vision understanding
- Research and self-hosting
thumb_upStrengths
- Broad Mistral model family
- Useful for multilingual and enterprise scenarios
- Includes coding, vision and open-weight branches
- Can be tiered by deployment and cost
infoLimitations
- Capabilities vary widely by model
- Self-hosting requires engineering effort
- Chinese quality needs separate evaluation
- Limits depend on Mistral documentation