Devstral Small 1.1
Mistral's efficiency model. Context window: 131K tokens.
Scores by test
Methodology →What you need to know
Devstral Small 1.1 is a low-cost utility model optimized for deterministic tasks rather than complex reasoning. Its primary value lies in its reliability for structured output, classification, and tool calling, all of which score 4/5. It also maintains high faithfulness and multilingual capabilities, making it a stable choice for data extraction and formatting.
The model struggles significantly with high-level cognitive tasks. It scores poorly (2/5) in agentic planning, strategic analysis, and creative problem solving, indicating it cannot independently manage complex workflows or nuanced personas. These deficits contribute to its low overall rank of 70 out of 71 models, suggesting it is not a general-purpose assistant.
At a blended cost of $0.250/MTok and a 131K context window, the model is priced for high-volume, simple operations. It provides a bargain for developers who need a lightweight engine to handle repetitive, structured data processing without paying for reasoning capabilities they do not require.
Use this model if you need an inexpensive tool for classification, structured data generation, or long-context retrieval. Skip this model if your application requires autonomous planning, complex strategic reasoning, or strict persona adherence.
Strengths — Top 3
Relative weaknesses — Bottom 3
Similar models