Use your Mistral Pro credits for AI Butler

Updated October 17, 2026 · Meowport Pro · About 7 minutes

Mistral's Pro plan comes with a monthly allowance that can be spent on its API as well as in its apps. Meowport recognizes Mistral's API and runs the Butler on it in full mode, so a Pro subscriber gets a capable European provider for the price they already pay. This guide shows the setup, what the allowance covers, and a few traps in Mistral's console.

Before you start. Prices, allowances and policies below are Mistral's, checked on October 17, 2026. They change without notice; confirm them on the provider's pages before you rely on them. You pay the provider directly: when an allowance runs out, requests may stop or cost extra. Keep your API key private, and if you suspect it leaked, remove it from Meowport and revoke it with the provider. You are responsible for the key, the requests it sends and the provider's terms. Meowport has not run a live test against Mistral's current models; the steps follow the provider's documentation.

Who this is for

You have Meowport Pro. AI Butler is a Pro feature; a custom model service only chooses which model answers.
You pay for Mistral's Pro plan (the plan behind its assistant, now called Vibe) and want its included allowance to cover the Butler. Mistral's free mode also allows API keys with a small monthly allowance, and pay-as-you-go works the same way in Meowport.
You are comfortable with each request's conversation and the pet context it needs being sent to Mistral under its terms, the same as with any custom provider.

Mistral's consumer terms do not say in so many words whether a personal plan's key may be used from a third-party app such as Meowport, while its commercial terms forbid buying, selling or transferring keys. If that matters to you, ask Mistral, or use a pay-as-you-go key, which Mistral's own help pages recommend for integrations.

What the allowance covers

Pro costs US$14.99 a month before tax and lists US$25.50 a month of API credits. The allowance is shared across Mistral Studio, the API and Vibe Code, and it is consumed first.
When the month's allowance is used up, usage can stop until the next billing period, unless you have turned on pay-as-you-go, in which case further usage draws from prepaid credits you add in the console. Rollover is not documented.
Pay-as-you-go is prepaid: you add credits, optionally with auto-recharge, and an organization-wide monthly spending limit can suspend access until the next month. With a zero balance the API fails with a 402 error.
Team and Enterprise plans do not document API credits; check with your administrator.

Step 1: Open the console and create a key

Sign in to Mistral's console at https://console.mistral.ai.
Open API Keys in the sidebar and choose Create new key. Give it a name, optionally an expiration, and copy it. The full key appears only once.
Keys belong to a workspace, not to a plan. A key created from the Vibe Code CLI page is tied to that tool's budget; for Meowport, create the key in the console as above, or a pay-as-you-go key if you have enabled billing.

Step 2: Connect it in Meowport

Open Meowport and go to Settings → AI Butler.
Choose Custom model API (marked Advanced).
Fill in the three fields:
- Base URL: https://api.mistral.ai/v1
- API Key: paste the key from Step 1.
- Model name: mistral-medium-latest is a good default and reads images; mistral-small-latest is cheaper; mistral-large-latest is the most capable.
Tap Fetch models to confirm the key works, and pick a model from the list if you prefer.
Tap Save model settings.

Mistral also offers an EU-only inference endpoint, https://api.eu.mistral.ai/v1, at a slightly higher price. Meowport treats it the same way.

How Meowport treats this provider

Mistral is one of the providers Meowport recognizes by its address. The Butler runs it in full mode: the model's plan is taken as it is and goes to your normal confirmation, and the model may look up more of your cat's records when a question needs them, within the bounds Meowport sets. The Before saving choice, action cards and undo from the chat work as with every provider.

To read a clinic report or a lab photo, choose a model that Mistral lists with image input, such as Medium or Large, and tap Test this model's image support once in the settings.

Step 3: Try it

Open the Butler and say something small, for example "Mochi weighed 4.2 kg this morning." The Butler drafts a weight record and, depending on your Before saving setting, saves it or asks first. Whatever you choose, it always asks before changing a record you already have, and it never deletes one. Anything it saves can be undone from the chat.

Costs and limits

Mistral's list prices at the time of writing: Small 4 at US$0.15 per million input tokens and US$0.60 per million output tokens, Large 3 at US$0.50 and US$1.50, Medium 3.5 at US$1.50 and US$7.50. An ordinary Butler request is a fraction of a cent on Small and still cheap on Medium.
Rate limits come in tiers. Free mode has the lowest; pay-as-you-go unlocks tier 1, and higher tiers follow your billed total. Adding credits alone does not raise the limits. The current numbers are in the console under Limits.
When the allowance is gone and pay-as-you-go is off, requests fail until the next period. Meowport shows Mistral's message and writes nothing.

Privacy and what is sent

With a custom model service, each request sends your message, any images you attached, and the pet records the Butler needs for that request to the provider, under the provider's terms and privacy policy. Mistral's privacy policy says inputs and outputs are kept for thirty rolling days to monitor abuse unless zero data retention is activated, which Mistral grants on request to pay-as-you-go customers. A separate toggle in the console opts your API traffic out of model improvement. Meowport's privacy policy describes the custom-provider path. Apple's on-device model stays available: switch back any time in the same settings. To stop using Mistral, clear the key in Meowport and delete it in the console.

Troubleshooting

Authentication error. The key was pasted incompletely, has expired or was deleted. Create a new one in the console.
402 error. The allowance is used up and no prepaid credits are available, or no payment method is on the account. Check the billing page in the console.
The model is not found. Model ids must match exactly; the -latest aliases are the safest. Tap Fetch models and pick from the list.
Image attachments are not offered. Choose a model with image input and run Test this model's image support.

Your responsibilities

Check the provider's pricing and terms yourself; this guide reflects October 17, 2026 and cannot follow later changes.
Keep the API key private. Do not paste it into chats, screenshots or shared notes; set an expiration if you can.
Watch your usage in the console. You, not Meowport, pay for what the key spends.
What you send is handled by Mistral under its own policies, outside Meowport's control.
AI can be wrong. Review what the Butler writes, and talk to your veterinarian before any medical decision.
A Meowport is a personal record, not a certificate.