Mistral Small 4: Intelligence, Performance & Price Analysis
Available providers
Model summary
Mistral Small 4 (Reasoning) is below average in intelligence, but well priced when comparing to other open weight models of similar size. It's also notably quick per task and highly concise. The model supports text and image input, outputs text, and has a 256k tokens context window.
Mistral Small 4 (Reasoning) scores 4 on the Artificial Analysis Intelligence Index, placing it below average among comparable models (median: 4). When evaluating the Intelligence Index, it generated 23.9M tokens, which is very concise in comparison to the median of 63.8M.
Pricing for Mistral Small 4 (Reasoning) is $0.15 per 1M input tokens (competitively priced, median: $0.30) and $0.60 per 1M output tokens (moderately priced, median: $0.80). On average, it costs $0.0399 per task to evaluate Mistral Small 4 (Reasoning) on the Intelligence Index.
At 1.7 minutes per Intelligence Index task, Mistral Small 4 (Reasoning) is notably quick compared with the median of 4.6 minutes for comparable models (lower is better).
| Input modality | Supports: text and image |
|---|---|
| Output modality | Supports: text |
| Context window | 256k ~384 A4 pages of size 12 Arial font |
| Total parameters | 119B |
| Active parameters | 6.5B Number of parameters active per token during inference |
| License | Apache 2.0 |
| Model weights | Hugging Face |
Metrics are compared against models of the same class:
- Non-reasoning models → compared only with other non-reasoning models
- Reasoning models → compared across both reasoning and non-reasoning
- Open weights models → compared only with other open weights models of the same size class:
- Tiny: ≤4B parameters
- Small: 4B–40B parameters
- Medium: 40B–150B parameters
- Large: >150B parameters
- Proprietary models → compared across proprietary and open weights models of the same price range, using a blended 3:1 input/output price ratio:
- <$0.15 per 1M tokens
- $0.15–$1 per 1M tokens
- >$1 per 1M tokens
Intelligence
Artificial Analysis Intelligence Index
Artificial Analysis Intelligence Index by Open Weights / Proprietary
Intelligence Index vs. Cost per Task
Capability Indexes
Incorporates 5 evaluations: AA-Briefcase v1.1, AA-Omniscience v1.1, HLE-Diamond, AutomationBench-AA, GDP.pdf · Higher is better
Incorporates 4 evaluations: AutomationBench-AA, AA-Briefcase v1.1, AA-Omniscience v1.1, GDP.pdf · Higher is better
Incorporates 5 evaluations: AA-Omniscience v1.1, AA-Briefcase v1.1, HLE-Diamond, GDP.pdf, AutomationBench-AA · Higher is better
Incorporates 5 evaluations: AA-Omniscience v1.1, AA-Briefcase v1.1, MLCR-AA, AutomationBench-AA, HLE-Diamond · Higher is better
Incorporates 7 evaluations: AA-Omniscience v1.1, CritPt v2 (Hard Subset), Terminal-Bench-Science 0.1, HLE-Diamond, Terminal-Bench 4.0, FrontierCode-AA 2.0, AA-Briefcase v1.1 · Higher is better
Incorporates 4 evaluations: AA-Omniscience v1.1, AA-Briefcase v1.1, HLE-Diamond, GDP.pdf · Higher is better
Evaluations
Agentic knowledge work, Elo
Knowledge
1 - hallucination rate
Agentic SaaS workflows
Agentic scientific research workflows in a terminal
Agentic software engineering in open-source repositories
Reasoning & knowledge
Professional document reasoning, All-pass
Agentic coding & terminal use
Physics reasoning
Quantitative analysis on spreadsheets & documents
Agentic real-world work tasks, Elo
Medical long context reasoning
Visual reasoning
Openness Index
Artificial Analysis Openness Index: Score
Cost
Cost per Intelligence Index Task
Cost to Run Artificial Analysis Intelligence Index
Pricing: Cache Hit, Input, and Output
Performance
Measured by time per task (minutes), output speed (tokens per second), time to first answer token and time to first token (seconds)
Time per Intelligence Index Task
Output Speed
Time to First Answer Token
Token Use
Output Tokens per Intelligence Index Task
Context Window
Context Window
Model Size (Open Weights Models Only)
Model Size: Total and Active Parameters
Further details
Frequently Asked Questions
Common questions about Mistral Small 4
Mistral Small 4 was released on March 16, 2026.
Mistral Small 4 was created by Mistral.
Mistral Small 4 has 2 effort levels: reasoning and non-reasoning. Mistral Small 4 scores 4 (reasoning) on the Artificial Analysis Intelligence Index. Mistral Small 4 (Reasoning) is below average among other open weight models of similar size (median: 4).
Mistral Small 4 takes 1.7 minutes (reasoning) per Intelligence Index task (based on Mistral's API; lower is better). Mistral Small 4 (Reasoning) is among the quickest compared to other open weight models of similar size (median: 4.6 minutes).
Mistral Small 4 has a time to first answer token of 13.41s (reasoning) and 1.61s (non-reasoning) (based on Mistral's API). Mistral Small 4 (Reasoning)'s time to first answer token is at the higher end compared to other open weight models of similar size (median: 8.18s).
Mistral Small 4 costs $0.15 per 1M input tokens (very competitive, median: $0.30) and $0.60 per 1M output tokens (better than average, median: $0.80), based on Mistral's API.
Mistral Small 4 costs $0.15 per 1M input tokens and $0.60 per 1M output tokens (based on Mistral's API). For a blended rate (7:2:1 cache hit/input/output ratio), this is $0.10 per 1M tokens. Pricing may vary by provider. Compare provider pricing
When evaluated on the Intelligence Index, Mistral Small 4 generated 23.9M output tokens (reasoning). Mistral Small 4 (Reasoning)'s output token count is very competitive compared to other open weight models of similar size (median: 63.8M).
Yes, Mistral Small 4 (Reasoning) is a reasoning model. It uses extended thinking or chain-of-thought reasoning to work through complex problems before providing an answer. No, Mistral Small 4 (Non-reasoning) is not a reasoning model. It provides direct responses without extended chain-of-thought reasoning.
Mistral Small 4 supports text and image input.
Mistral Small 4 supports text output.
Yes, Mistral Small 4 supports image input and can analyze, describe, and answer questions about images.
Yes, Mistral Small 4 is multimodal. It can process text and image input and generate text output.
Mistral Small 4 has a context window of 256k tokens. This determines how much text and conversation history the model can process in a single request.
Yes, Mistral Small 4 is open weights. The model weights are publicly available and can be downloaded for self-hosting.
Mistral Small 4 has 119B parameters (6.5B active).
Mistral Small 4 is a Mixture of Experts (MoE) model with 119B total parameters, but only 6.5B active parameters are used during inference.
Mistral Small 4 is released under the Apache 2.0 license. This license allows commercial use. View license
Mistral Small 4 achieves a score of 4 (reasoning) on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across knowledge work, coding, science, and general knowledge.
Yes, Mistral Small 4 is available via API through 1 provider. Compare API providers
Mistral Small 4 is available through 1 API provider. Compare providers