Mistral AI logo

Mistral Small 4

mistral/mistral-small-2603

Fast Mistral production model for chat, extraction, and cost-sensitive agents

Source-linkedmistral-smallOpen weightsReleased 2026-03-16
Report data
Context window256K
Maximum output256K
Provider offers4
Knowledge cutoff2025-06
Capabilities

What this model supports

ReasoningYes
Tool callingYes
Structured outputNo
AttachmentsYes
Temperature controlYes
Open weightsYes
Creator
Mistral AI
Family
mistral-small
Modalities
Image, Text
License
Not documented
Release date
2026-03-16
Model ID
mistral/mistral-small-2603
Market availability

Provider pricing

Report price →
Providers offering Mistral Small 4
ProviderProvider model IDContextInput / 1MOutput / 1MCached / 1MFeaturesAll pricing
Pioneermistralai/Mistral-Small-4-119B-2603262.144K$0.150$0.600$0.015R T ·
View
{
  "cache_read": 0.015,
  "cache_write": 0.15,
  "input": 0.15,
  "output": 0.6
}
Mistralmistral-small-2603256K$0.150$0.600R T ·
View
{
  "input": 0.15,
  "output": 0.6
}
OpenRoutermistralai/mistral-small-2603262.144K$0.150$0.600$0.015R T S
View
{
  "cache_read": 0.015,
  "input": 0.15,
  "output": 0.6
}
Venice AImistral-small-2603256K$0.188$0.750R T S
View
{
  "input": 0.1875,
  "output": 0.75
}
Evidence

Published benchmark results

Artificial Analysis Coding Index vs top 8 models
GPT-5-Codex
38.9
Step 3.5 Flash 2603
34.6
North Mini Code
33.4
Gemini 2.5 Pro
32.0
Step 3.5 Flash
31.6
GLM-4.6
29.5
Qwen3 Max
26.4
Mistral Small 4
24.3
SciCode vs top 8 models
Fugu
60.1
Fugu Ultra
58.7
Qwen3.7 Max
53.5
Gemini 2.5 Pro
42.8
GPT-5-Codex
40.9
Step 3.5 Flash
40.4
Gemini 2.5 Flash
39.4
Mistral Small 4
38.0
Terminal-Bench Hard vs top 8 models
GPT-5-Codex
37.9
Step 3.5 Flash 2603
32.6
Step 3.5 Flash
27.3
Gemini 2.5 Pro
26.5
GLM-4.6
25.0
GLM-4.5
22.0
GLM-4.5-Air
20.5
Mistral Small 4
17.4