N
NexusRoute
Back to Models

Mistral Medium 3

Mistralmid

Mistral's balanced mid-tier model offering 8x cheaper pricing than frontier models while retaining strong quality. 128K context with good multilingual and coding capabilities. Designed as the everyday workhorse for Mistral API users.

Released 2026-01-15Knowledge cutoff: 2025-10
Medium confidence|Updated 88d ago|82% source confidence

Specifications

Context Window

128K tokens

Max Output

16.4K tokens

Input Price

$0.400 / 1M tokens

Output Price

$2.00 / 1M tokens

Latency Tier

Fast (speed score: 8/10)

Capability Profile

Cost Efficiency
8.5/10
Speed
8/10
Structured Output
7.5/10
Instruction Following
7.5/10
Conversational
7.5/10
Reasoning
7/10
Coding
7/10
Factuality
7/10
Tool Use
7/10
Safety & Enterprise
7/10
Long Context
6.5/10
Creativity
6.5/10
Multimodal
5.5/10

Feature Support

Vision Yes
Audio In No
Audio Out No
Video No
Image Generation No
Image Editing No
Function Calling Yes
JSON Mode Yes
Structured Output Yes
Streaming Yes
Reasoning No
Realtime No
Computer Use No
Web Search No

Best Use Cases

Cost-effective multilingual production applications
General-purpose assistant deployments on the Mistral platform
Function calling and structured extraction at mid-tier pricing
European enterprise applications with data residency needs
Batch processing where Mistral Large 3 quality is overkill

Not Ideal For

Complex coding requiring frontier quality
Tasks needing the longest context windows
Audio or video processing
Safety-critical applications where Anthropic-level alignment is needed

Strengths

Strong cost-to-quality ratio — good enough for most tasks at mid-tier pricing
Inherited multilingual strength from the Mistral family
EU data residency available
Vision support at a mid-tier price point
Reliable function calling and JSON mode

Weaknesses

Coding quality below GPT-5.4-mini and Claude Sonnet 4.6
128K context is modest by current standards
Less widely adopted than competitors — smaller community
Creative output is competent but unremarkable
Vision capabilities are less advanced than Gemini or GPT

Edge Cases & Notes

EU data residency option makes it valuable for European compliance scenarios
Vision quality on complex images is below Gemini 2.5 Flash
Good default choice for Mistral platform users who don't need frontier quality

Provider Notes

Available through Mistral's La Plateforme and Azure AI. EU data residency available. Consider GPT-5.4-mini or Gemini 2.5 Flash for broader capability at similar pricing.

Benchmarks

MMLU83.5%
HumanEval82%
Arena Elo1260

Benchmark Notes

MMLU-Pro 83.5%. Solid mid-tier performance. Multilingual benchmarks are its strongest area relative to its tier. Good value on cost-adjusted metrics.

Research Meta

Last Evaluated

2026-03-01

Source Confidence

82%

Evaluation Method

LMSYS Arena, MMLU-Pro, multilingual evaluation, cost-quality analysis

Needs Re-evaluation

No

Sources

  • Mistral Medium 3 announcement
  • LMSYS Chatbot Arena
  • Mistral AI benchmarks

Continue exploring

Route a prompt

See how Mistral Medium 3 ranks

Compare models

Side-by-side analysis

Browse registry

Explore all 24 models