ModelBench
Mistral AI logo
Models/Mistral AI

Mistral AI: Mistral Small 4

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

Source-linked mistral-small Open weights Released 2026-03-16
API record Report
InputT
OutputT
Input price$0.15/M
Output price$0.6/M
Context256K
Max output256K
Providers12
Inference availability

Providers

Provider-specific identifiers, limits, and listed prices per million tokens. Every row links back to the provider's own documentation.

Report a price
Providers offering Mistral Small 4
ProviderProvider model IDContextMax outputInputOutputCache readCapabilitiesDocs
Cortecs mistral-small-2603 262.144K 256K $0.143 $0.568 $0.014 ReasoningToolsJSON Docs ↗
Eden AI mistral/mistral-small-2603 262.144K 256K $0.15 $0.6 $0.015 ReasoningToolsJSON Docs ↗
Pioneer mistralai/Mistral-Small-4-119B-2603 32K 32K $0.15 $0.6 $0.015 ReasoningTools Docs ↗
OpenRouter mistralai/mistral-small-2603 262.144K 209.715K $0.15 $0.6 $0.015 ReasoningToolsJSON Docs ↗
Mistral mistral-small-2603 256K 256K $0.15 $0.6 — ReasoningTools Docs ↗
Opper mistral/mistral-small-2603 256K 256K $0.15 $0.6 — ReasoningTools Docs ↗
EmpirioLabs AI mistral-small-4 256K 65.536K $0.15 $0.6 $0.15 ToolsJSON Docs ↗
Kilo Gateway mistralai/mistral-small-2603 262.144K 209.715K $0.15 $0.6 $0.015 ReasoningToolsJSON Docs ↗
Requesty mistral-small-2603 256K 256K $0.165 $0.66 $0.165 ReasoningToolsJSON Docs ↗
Venice AI mistral-small-2603 256K 65.536K $0.188 $0.75 — ReasoningToolsJSON Docs ↗
AKI.IO mistral4-119b 262.144K 81.92K $0.2 $0.6 — ReasoningTools Docs ↗
Infomaniak mistralai/Mistral-Small-4-119B-2603 256K 256K $0.25 $0.93 — ReasoningTools Docs ↗

Capability badges appear only where the provider catalog explicitly lists support. A blank cell means the source is silent, not that the feature is absent.

Listed rates

Price across providers

Input price per million tokens as published by each provider. Bars are drawn from listed rates only — no traffic weighting, since the catalog observes no requests.

Lowest input $0.143/M

Across 12 priced providers

Median input $0.15/M

Midpoint of listed rates

Highest input $0.25/M

1.7× the lowest listed rate

Output range $0.568 – $0.93

Per million output tokens

Cortecs $0.143/MLowest
Eden AI $0.15/M
Pioneer $0.15/M
OpenRouter $0.15/M
Mistral $0.15/M
Opper $0.15/M
EmpirioLabs AI $0.15/M
Kilo Gateway $0.15/M
Requesty $0.165/M
Venice AI $0.188/M
AKI.IO $0.2/M
Infomaniak $0.25/M
Cost calculator

Estimate a workload

$0.00
Excludes taxes, non-token charges, and tiered discounts.

Context limits also differ by provider, from 32K to 262.144K tokens. Compare the provider table above before choosing on price alone.

Specification

Capabilities

Recorded from the source catalog and provider listings.

✓ Reasoning Yes
✓ Tool calling Yes
? Structured output Unknown
✓ Attachments Yes
✓ Vision input Yes
✓ Open weights Yes
Creator
Mistral AI
Model family
mistral-small
Knowledge cutoff
2025-06
License
Not documented
Release date
2026-03-16
Model ID
mistral/mistral-small-2603

Weights: Hugging Face ↗

Published evaluations

Benchmarks

Every result stays attached to its source, version, metric and harness. Scores from different versions are never merged, and the profile below plots each benchmark against its own population rather than on a shared scale.

Benchmark registry
Benchmark profile
SciCodeSciCode AA Coding IndexArtificial Analysis Coding Index Terminal-Bench…Terminal-Bench Hard This model — SciCode: 38 percent correct (68.5th percentile) This model — Artificial Analysis Coding Index: 24.3 index (67.3th percentile) This model — Terminal-Bench Hard: 17.4 success rate (52.3th percentile)

Each axis is this model's percentile among the 3 benchmarks it has published results for, measured against every other model with a score on that same benchmark. Percentiles are used because benchmarks do not share a scale — a 60 on one is not a 60 on another. Hover any point for the raw score.

SciCode percent correct · 2026-06-01 Source ↗
SciCode #9 of 27
Artificial Analysis Coding Index index · 2026-06-01 Source ↗
Artificial Analysis Coding Index #9 of 26
Terminal-Bench Hard success rate · 2026-06-01 Source ↗
Terminal-Bench Hard #11 of 22

Each strip shows every published score for that benchmark, with this model marked. The lighter shape behind the ticks is the density of results, and the dashed line is the median.

Cost against capability

Price and performance

Listed input price plotted against SciCode, the benchmark with the widest published coverage that this model appears in.

0 23 45 $0.1 $1 $10 Qwen3-Coder 30B-A3B Instruct — $0.45/M, score 27.8 Qwen3 Max — $1.2/M, score 38.3 DeepSeek-R1 — $0.7/M, score 35.7 Gemini 2.5 Flash — $0.3/M, score 39.4 Gemini 2.5 Flash-Lite — $0.1/M, score 19.3 Gemini 2.5 Pro — $1.25/M, score 42.8 Llama-3.3-70B-Instruct — $0.1/M, score 26.0 Devstral 2 — $0.4/M, score 33.1 Mistral Large 2.1 — $2/M, score 29.2 Mistral Large 3 — $0.5/M, score 36.2 Mistral Medium 3 — $0.4/M, score 33.1 Mistral Small 4 — $0.15/M, score 38.0 GPT-4 Turbo — $10/M, score 31.9 GPT-4o (2024-05-13) — $5/M, score 30.9 GPT-4o (2024-08-06) — $2.5/M, score 33.1 GPT-4o (2024-11-20) — $2.5/M, score 33.3 GPT-4o mini — $0.15/M, score 22.9 GPT-5-Codex — $1.1/M, score 40.9 Sonar — $1/M, score 22.9 Sonar Pro — $3/M, score 22.6 Step 3.5 Flash — $0.1/M, score 40.4 Step 3.5 Flash 2603 — $0.1/M, score 38.5 Step 3.7 Flash — $0.185/M, score 40.0 GLM-4.5 — $0.6/M, score 34.8 GLM-4.5-Air — $0.2/M, score 30.6 GLM-4.5V — $0.6/M, score 22.1 GLM-4.6 — $0.6/M, score 38.4 Mistral Small 4 Input price per million tokens (log scale) Percent correct

The stepped line is the efficient frontier: at each price, the best score available for that money or less. A model sitting on it is not being beaten by anything cheaper. Price is log-scaled because listed rates span four orders of magnitude. Only models with both a listed price and a score on this benchmark can appear.

Catalog activity

Change log

Field-level changes detected between successful source imports.

Full change log
No changes recorded

This record has not changed within the retained import history.

Provenance

Sources & verification

Every figure on this page traces back to one of these records.

Methodology
Last verified
Sep 11, 2026
Status
Source-linked
Confidence
Medium
Catalog source
Models.dev
Models.devcatalog ↗Cortecs documentationprovider documentation ↗Eden AI documentationprovider documentation ↗Pioneer documentationprovider documentation ↗OpenRouter documentationprovider documentation ↗Mistral documentationprovider documentation ↗Opper documentationprovider documentation ↗EmpirioLabs AI documentationprovider documentation ↗Kilo Gateway documentationprovider documentation ↗Requesty documentationprovider documentation ↗Venice AI documentationprovider documentation ↗AKI.IO documentationprovider documentation ↗
Public API

Use this record

Fetch the complete source-linked model record. No key, no account, no rate-limited tier.

API documentation
Endpoint
GET https://model.kyssta.lol/api/v1/models/mistral/mistral-small-2603
curl
curl "https://model.kyssta.lol/api/v1/models/mistral/mistral-small-2603"
Common questions

Frequently asked questions

Answered directly from the stored record — nothing here is generated beyond the catalog's own fields.

What is Mistral Small 4?

Fast Mistral production model for chat, extraction, and cost-sensitive agents. It is published by Mistral AI and catalogued here from Models.dev.

How much does Mistral Small 4 cost?

Listed input pricing starts at $0.143 per million tokens from Cortecs, rising to $0.25 across 12 listed providers.

What is the context length of Mistral Small 4?

Mistral Small 4 accepts up to 256K tokens of context and returns up to 256K output tokens.

Does Mistral Small 4 support tool calling and structured output?

Provider catalogs list support for tool calling, reasoning, and image input.

Which providers serve Mistral Small 4?

12 providers list this model: Cortecs, Eden AI, Pioneer, OpenRouter, Mistral, Opper and 6 more.

Are the weights for Mistral Small 4 open?

Yes. The weights are published and downloadable from Hugging Face.

When was Mistral Small 4 released?

The catalog records a release date of 2026-03-16, last verified Sep 11, 2026.

More models from Mistral AI

View all →
Mistral: Mistral Small 3

Mistral Small 3 is a 24B-parameter language model optimized for low-latency performance across common AI tasks. Released under the Apache 2.0 license, it features both pre-trained and instruction-tuned versions designed...

32.768K ctx $0.05/M in 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...

262.144K ctx $0.15/M in Mistral: Ministral 3 3B 2512

The smallest model in the Ministral 3 family, Ministral 3 3B is a powerful, efficient tiny language model with vision capabilities.

131.072K ctx $0.1/M in Mistral: Saba

Mistral Saba is a 24B-parameter language model specifically designed for the Middle East and South Asia, delivering accurate and contextually relevant responses while maintaining efficient performance. Trained on curated regional...

32.768K ctx $0.2/M in