Mistral: Mistral Small 4

Mistral: Mistral Small 4

mistralai · Released Mar 16, 2026 Efficient
Intelligence #172 / 583
44.7 Our Score
Speed #62 / 276
159.1 tokens / sec
Input #250 / 583
$0.150 per 1M tokens
Output #272 / 583
$0.600 per 1M tokens
Context #110 / 583
262,144 tokens

Analysis Summary

Mistral Small 4 is a compact model from Mistral AI with a 262K context window, vision support, tool use, and function calling. Its benchmark scores place it in the lower-mid tier for reasoning and coding, but its multimodal capability and low price point make it a practical option for teams that need a versatile, affordable workhorse rather than frontier-level depth.

For businesses, it suits lighter content generation, document summarisation, and structured output tasks where cost efficiency matters more than peak reasoning. The vision modality adds flexibility for image-aware workflows. It is not the right choice for complex agentic pipelines or demanding software engineering tasks, where its coding and agentic scores fall short.

At $0.15 input and $0.60 output per million tokens, it competes well on price. Teams running high-volume, moderate-complexity tasks, such as content drafting, classification, or basic tool-augmented workflows, will find it a cost-effective fit.

Assessed June 17, 2026

Editorial notes

Mistral Small 4 is a lightweight multimodal model with tool use, function calling, and vision support at very competitive pricing, though its reasoning and coding benchmarks sit well below frontier tier.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence2.2Technical2.4Value8Content5.5
Intelligence 2.2/10
Technical 2.4/10
Content 5.5/10
Value 8/10

How Mistral: Mistral Small 4 compares

Mistral: Mistral Small 4 ranks #131 of 380 AI models we track for overall intelligence, #148 of 292 for agentic tasks. Its 262K-token context window is larger than 81% of the models we list. At $0.15 per million input tokens it is cheaper than 57% of comparable models.

About Mistral: Mistral Small 4

Mistral Small 4 is the next major release in the Mistral Small family, unifying the capabilities of several flagship Mistral models into a single system. It combines strong reasoning from..

Capabilities

Tool Use Function Calling Vision

Performance Indices

Source: Artificial Analysis

20.8 Intelligence Index
29.3 Agentic Index

Benchmark Scores

Intelligence

GPQA Diamond 76.9% Graduate-level scientific reasoning
HLE 9.5% Humanity's Last Exam
SciCode 38% Scientific computing

Technical

TerminalBench Hard 17.4% Agentic terminal tasks
τ²-Bench 41.2% Conversational agent benchmark

Content

IFBench 48.2% Instruction following
LCR 44.7% Long-context reasoning

Benchmark data from Artificial Analysis and Hugging Face

How does Mistral: Mistral Small 4 stack up?

Compare side-by-side with other efficient models.

Compare Models

Model Information

OpenRouter ID mistralai/mistral-small-2603
Providermistralai
Model FamilyMistral
Release Date March 16, 2026
Context Length262,144 tokens
Status Active

Pricing

Token Type Cost per 1M tokens Cost per 1K tokens
Input $0.15 $0.000150
Output $0.60 $0.000600

Live Performance

Live endpoint metrics, refreshed every 30 minutes.

99.9%
Avg Uptime
403ms
Best Latency (TTFT)
113 tok/s
Best Throughput
2/2
Active Endpoints
Available via: Mistral, Venice

Leaderboard Categories

Frequently asked questions about Mistral: Mistral Small 4

How much does Mistral: Mistral Small 4 cost?

Mistral: Mistral Small 4 costs $0.15 per million input tokens and $0.60 per million output tokens.

What is the context window of Mistral: Mistral Small 4?

Mistral: Mistral Small 4 has a context window of 262,144 tokens (262K).

What can Mistral: Mistral Small 4 do?

Mistral: Mistral Small 4 supports image/vision input, tool use, and function calling.

Who created Mistral: Mistral Small 4?

Mistral: Mistral Small 4 is developed by Mistral and was released on March 16, 2026.