Mistral AI
Mistral API β European-sovereign models, no separate account.
mistral-large, mistral-small, codestral, and mixtral β fast, license-friendly European LLMs through one OpenAI-compatible endpoint.
Latest models as of 2026-05-23
See all models βmistral-medium-2-latest balanced Production sweet spot. Lower cost than Large with similar capability for most tasks.
codestral-2-latest specialized Code-specialized model with strong fill-in-the-middle. Used in many IDE integrations.
mistral-large-2-latest flagship Mistral's top closed-weight model. EU-hosted option available for GDPR-strict workloads.
Why use JJAPI for this API
European data residency option
We can route your Mistral requests to EU-hosted endpoints for GDPR-sensitive workloads.
Code completion via FIM
Codestral supports fill-in-the-middle, perfect for editor integrations like Continue and Cody.
Permissive licensing
Mixtral models are Apache-2.0 β safe for commercial use and self-hosting.
OpenAI-compatible
Drop-in replacement for Mistral's own API with broader rate limits.
from openai import OpenAI
client = OpenAI(
base_url="https://api.jjapi.net/v1",
api_key="sk-jjapi-...",
)
r = client.chat.completions.create(
model="mistral-medium-2-latest", # current as of 2026-05-23
messages=[{"role": "user", "content": "Hi"}],
)
print(r.choices[0].message.content) Frequently asked
Can I use Mistral for GDPR-strict workloads?
Yes β request EU-only routing in your console settings and your traffic will only hit Mistral's Paris infrastructure.
Does codestral support FIM (fill-in-the-middle)?
Yes β pass 'prompt' and 'suffix' params to /v1/completions and codestral fills the gap. Useful for IDE-style autocomplete.
What's the difference between mistral-large and mixtral?
mistral-large is closed-weight and Mistral's strongest model. mixtral-8x22b is open-weight, slightly weaker but Apache-2.0 licensed for self-hosting.