llmproviders.ai
← All providers

Mistral pricing

verified 2026-09-02

European lab shipping efficient open and commercial models, popular where data residency and price both matter.

data residency & retention
from Mistral's own policies · verified 2026-09-02
Headquarters
🇫🇷 France (EU)
Where inference runs
EU, US
By default, data is hosted in the European Union, with an optional explicit US API endpoint; temporary transfers to subprocessors outside the EU are possible per its Trust Center list.
Region pinning
Yes, self-serve
EU hosting is the default and the US endpoint is an explicit opt-in; enterprise customers can additionally deactivate features involving data transfers outside the EU.
Prompt retention
Retained · 30 days
API inputs and outputs are kept 30 rolling days for abuse monitoring, then deleted, unless Zero Data Retention is activated.
Trains on API data
No
Pay-as-you-go API data is not used for training by default; free/experimental-tier data may be used unless you opt out.
Zero data retention
Available
ZDR is available on pay-as-you-go for stateless endpoints via a request approved at Mistral's discretion; it does not cover stateful products (agents, files, batch).
current models · $ per 1M tokens
ModelInput /1MCached inOutput /1MContext
Mistral Small 4
Mistral Small
$0.15—$0.60—
Mistral Medium 3.5
Mistral Medium
$1.50—$7.50—
Mistral Large 3
Mistral Large
$0.50—$1.50—
Ministral 3 14B
Ministral
$0.20—$0.20—
Ministral 3 8B
Ministral
$0.15—$0.15—
Ministral 3 3B
Ministral
$0.10—$0.10—
Codestral
Codestral
$0.30—$0.90—
Mistral totals · 1M in + 1M out
Ministral 3 3B
$0.2000
Ministral 3 8B
$0.3000
Ministral 3 14B
$0.4000
Mistral Small 4
$0.7500
Codestral
$1.20
Mistral Large 3
$2.00
Mistral Medium 3.5
$9.00

Cost of 1M input + 1M output tokens. Bar length is square-root scaled so cheap models stay visible next to premium ones.