Mistral AI

European lab with an open-weight lineage and EU-resident hosting

mistral.ai docs.mistral.ai

Overall in llm apis

1of 18 · score 76.6

Data coverage

85%

23 of 27 fields sourced

Best for

•EU data-residency requirements that rule out US-hosted inference
•On-premise deployment with a commercial support contract
•Teams that want an open-weight escape hatch under the hosted API

Strong on

+Notice period 180 days 1/7
+$/M cache read $0.05/M tok 2/12
+MCP support Partial 3/18

Weak on

−Context window 256,000 tokens 10/14
Better: Anthropic: 1,000,000 tokens

vs Groq

•Ahead on Effort control: Partial vs No
•Ahead on MCP support: Partial vs No
•Behind on Zero retention: Partial vs Yes
•Behind on $/M input: $0.5/M tok vs $0.05/M tok

How Mistral AI scores among 18 llm apis23 of 27 fields known

Capability7

Flagship model

Mistral Large 3 (mistral-large-3-25-12)

docs.mistral.ai · 2026-07-30

Context window

256,000 tokens

10/14

Leader: Anthropic · 1,000,000 tokens

docs.mistral.ai · 2026-07-30

Max output

—

Image input1st

Yes

1/18

docs.mistral.ai · 2026-07-30

Audio in/out1st

Partial

1/18

docs.mistral.ai · 2026-07-30

Open weights1st

Yes

1/18

docs.mistral.ai · 2026-07-30

OpenAI-compat API1st

Yes

1/18

docs.mistral.ai · 2026-07-30

Performance2

TTFT p50

—

Output tok/s

—

Pricing4

$/M input

$0.5/M tok

5/13

Leader: Groq · $0.05/M tok

mistral.ai · 2026-11-06

$/M output

$1.5/M tok

5/13

Leader: Fireworks AI · $0.05/M tok

mistral.ai · 2026-11-06

$/M cache read2nd

$0.05/M tok

2/12

Leader: DeepSeek · $0.0028/M tok

mistral.ai · 2026-07-30

Batch discount1st

50%

1/14

mistral.ai · 2026-07-30

Agentic6

Tool use1st

Yes

1/18

docs.mistral.ai · 2026-07-30

Schema output1st

Yes

1/18

· 2026-01-15

Effort control

Partial

5/18

Leader: Anthropic · Yes

· 2026-01-15

Computer use

5/18

Leader: Anthropic · Yes

· 2026-01-15

MCP support3rd

Partial

3/18

Leader: Anthropic · Yes

· 2026-01-15

Cache TTL

—

Governance & continuity6

Continuity policy

Dated model IDs (the -2411 style suffix) are stable and documented, and Mistral maintains a legacy-model page listing deprecation and retirement dates. The stronger guarantee is structural: for the Apache-licensed models you can keep serving the same weights yourself after the endpoint goes away.

· 2026-01-15

Notice period1st

180 days

1/7

docs.mistral.ai · 2026-11-07

Pinnable versions1st

Yes

1/18

docs.mistral.ai · 2026-01-15

Zero retention

Partial

7/18