Pixtral 12B, run in Paris by Scaleway.
Pixtral 12B is an open-weight model by Mistral AI built to read images, released under the Apache-2.0 licence. On Evanesca, Scaleway runs it in Paris: your prompts are processed in France, and Evanesca keeps no copy of them.
Checked on 20 Oct 2026
Pixtral 12B
by Mistral AI · Apache-2.0
- Input / M
- €0.336
- Output / M
- €0.336
- Quick question
- €0.0002
Prices per million tokens, VAT included. Quick question: 300 tokens in, 400 out.
The model at a glance
- Maker
- Mistral AI
- Licence
- Apache-2.0
- Size
- 12B parameters
- Context
- 128k tokens as the operator announces it
- Type
- Vision
- Tool calling
- Yes measured by a real call
- Reasoning
- —
- Run by
- Scaleway Scaleway SAS, Paris — iliad group, no US parent
- Where it runs
- Paris, France
- Checks
- 5/6 · Security ✕ Passed: Licence, Provenance, GDPR, Benchmarks and Live test
- Price
- €0.336 per million input tokens · €0.336 per million output tokens VAT included; reasoning tokens count as output
- Open weights
- Yes published by the maker: an operator can run the model
Where your prompts go
Here is the route of a message sent to Pixtral 12B, end to end. No call to the model leaves Europe: not as a backup, not when servers are busy.
- From your browser to EvanescaEvanesca passes your message straight on to Scaleway. It stores neither the prompt nor the answer: both stay in memory only while the answer is written.
- Scaleway, ParisScaleway runs Pixtral 12B in Paris, France. It is a French company with no US parent: out of the CLOUD Act's direct reach. Scaleway may keep and read requests that cause errors or look abusive, for up to 14 days (Scaleway's data privacy page).
- Back to you, with its passportThe answer comes back with its passport: model, operator, country, and what was kept. Your history lives in your browser; about this message, Evanesca keeps only billing data — tokens and cost.
What it costs
€0.336 per million input tokens and €0.336 per million output tokens, VAT included. Three typical exchanges on Pixtral 12B:
| Example | Tokens in | Tokens out | Cost |
|---|---|---|---|
| A quick question | 300 | 400 | €0.0002 |
| A long document to summarise | 15,000 | 800 | €0.0053 |
| A message late in a long conversation | 40,000 | 2,000 | €0.0141 |
€2 of credit covers about 8,500 quick questions on this model. Reasoning tokens count as output, so a model that thinks first can cost more than these examples. The exact cost of each answer shows under it. All prices.
Licence
Mistral AI publishes Pixtral 12B under the Apache-2.0 licence. Apache-2.0 is a permissive open-source licence: it allows commercial use.
The official model card is the reference.
Similar models
Same maker first, then the same type.
EU region of Mistral AIMistral Large 4
Mistral AI · Text + images · checks 4/6
€2.5133 · €7.7246 / M tokens
EU region of Mistral AIMistral Small 4
Mistral AI · Text + images · checks 5/6
€0.2772 · €1.1088 / M tokens
Paris, FranceMistral Small 3.2
Mistral AI · Text + images · checks 5/6
€0.252 · €0.588 / M tokens
Frequently asked questions
Where does Pixtral 12B run on Evanesca?
In Paris, France, on Scaleway's servers. Scaleway is a French company with no US parent, so your prompts stay out of the CLOUD Act's direct reach. No call to the model leaves Europe: not as a backup, not when servers are busy.
Does Evanesca keep what I send to Pixtral 12B?
No. The prompt and the answer stay in memory only while the answer is written, and your history lives in your browser; about your messages, Evanesca keeps only billing data — tokens and cost. Scaleway may keep and read requests that cause errors or look abusive, for up to 14 days (Scaleway's data privacy page).
How much does Pixtral 12B cost?
€0.336 per million input tokens and €0.336 per million output tokens, VAT included. A quick question — 300 tokens in, 400 out — costs about €0.0002. There is no subscription: credit starts at €2.
Can Pixtral 12B call tools?
Yes: a real call through Scaleway returned proper tool calls (function calling).
Which licence is Pixtral 12B under?
Apache-2.0. Apache-2.0 is a permissive open-source licence: it allows commercial use.
Which checks has Pixtral 12B passed?
5 of 6: Licence, Provenance, GDPR, Benchmarks and Live test. Security: run on 20 Oct 2026 without meeting the pass mark. Its passport shows the same stamps; how each check works.
Does Mistral AI see what I type?
No. Mistral AI published the weights, and Scaleway runs them in France: your prompts go to Scaleway's servers, not to Mistral AI.