Sources
Every source, and when it was read
All sources were read on 20 November 2026. The argument and the quotes come from A European brain for a sovereign website by Dinis Cruz. Mistral’s facts were checked on Mistral’s own pages that day by this site’s research agent, and are cited to those pages. Terms and prices change: these are as the pages showed that day.
The facts, one by one
- Mistral AI was founded in April 2023, and describes itself as “a European leader” in AI. (Source: Mistral AI, About; as read on 20 November 2026.)
- Mistral offers an OpenAI-compatible endpoint at
https://api.mistral.ai/openai/v1, in public preview; its standard/v1/chat/completionsuses Mistral’s own, similar format. Ollama and OpenRouter speak the same OpenAI-style API. (Source: Mistral docs, Chat completion and Ollama docs, OpenAI compatibility and OpenRouter, Quickstart; as read on 20 November 2026.) - Ministral 3 was announced on 2 December 2025 in 3B, 8B and 14B sizes, for edge and local use, each with image understanding, under Apache 2.0. (Source: Mistral AI, Mistral 3; as read on 20 November 2026.)
- By default, data is hosted in the EEA. (Source: Mistral help centre, Where do you store my data; as read on 20 November 2026.)
- Hosting is not processing: Mistral does not commit to an inference location on the global endpoint,
api.mistral.ai. Processing in the EU needs the EU endpoint,api.eu.mistral.ai, billed at 1.1 times list price, without Agents, Batch or the Files API, and serving only the models hosted there. (Source: Mistral docs, Regional inference; as read on 20 November 2026.) - Mistral’s regional endpoints were announced as generally available on 11 August 2026. (Source: Mistral AI, Regional inference announcement; as read on 20 November 2026.)
- Mistral’s inference page says “Data sent through our API isn’t used for model training”, but its help centre says that in Free mode it may use API inputs and outputs for training unless you opt out (a toggle in the Admin panel, under Privacy), and that Labs and Public Preview models may be used for training whatever the plan. (Source: Mistral AI, Cloud inference and Mistral help centre, Do you use my data to train and Mistral help centre, Opting out of training; as read on 20 November 2026.)
- Ministral 3 8B is listed at 0.15 dollars per million tokens in and 0.15 out. (Source: Mistral docs, Ministral 3 8B and Mistral docs, Pricing; as read on 20 November 2026.)
- Mistral Small 4 is an Apache 2.0 model with 119 billion parameters, 6.5 billion active, listed at 0.15 dollars per million input tokens and 0.60 per million output tokens. (Source: Mistral docs, Mistral Small 4 and Mistral docs, Pricing; as read on 20 November 2026.)
- Mistral offers function calling (Ministral 3 and Mistral Small 4 among the models), and connectors that register remote MCP servers as tools, in public preview. (Source: Mistral docs, Function calling and Mistral docs, Connectors; as read on 20 November 2026.)
- API inputs and outputs are kept for 30 rolling days to monitor abuse, unless zero data retention is on; Agents API data is kept until the account ends. Zero data retention is only for pay-as-you-go, only for stateless calls, must be requested and may be refused, and does not cover Labs or Public Preview models. (Source: Mistral AI, Privacy policy and Mistral help centre, Zero data retention and Mistral docs, Zero data retention; as read on 20 November 2026.)
- Some features may move data temporarily outside the EU to listed subprocessors; Enterprise customers can switch some of these off. (Source: Mistral help centre, Where do you store my data; as read on 20 November 2026.)
- Mistral says it “complies with both SOC 2 Type II and ISO 27001/27701 frameworks”. (Source: Mistral help centre, SOC 2 and ISO 27001; as read on 20 November 2026.)
- Ministral 3, Mistral Small 4 and Voxtral Small are labelled Apache 2.0. Mistral Large 4 is under a modified MIT licence: its announcement says a company offering it as a model-as-a-service needs a separate commercial licence, while Mistral’s help centre describes the condition as a monthly revenue threshold of 20 million dollars. The pages disagree; read the licence file itself before relying on either. (Source: Mistral docs, Ministral 3 8B and Mistral AI, Mistral Large 4 and Mistral help centre, Licences of the open models; as read on 20 November 2026.)
- Mistral lists strategic, ecosystem and integration partners and has a public “Partner with Mistral” expression-of-interest form; its ambassador programme takes rolling applications and offers free API credits, for at least six months’ commitment. (Source: Mistral AI, Partners and Mistral docs, Ambassadors; as read on 20 November 2026.)
- On 18 October 2026, Mistral’s free mode was one of four providers that passed the Newsroom’s two tests: free to start, and callable from a web page with the reader’s own key. (Source: Free models, your own key (SGit Newsroom); as read on 20 November 2026.)
- On 11 November 2026, at Web Summit, the Swarm ran 216 model calls on four open-weight models (DeepSeek, Llama, Mistral and Qwen), through OpenRouter, for 0.068 dollars. Which Mistral model, and its share of the calls, was not recorded. (Source: Pennies against dollars (SGit Newsroom); as read on 20 November 2026.)
- The article’s estimate, not measured: the 3B and 8B, compressed to about 2 to 5GB, should run on a 16GB Raspberry Pi 5, the 8B slowly; on a tablet, the 3B is the realistic one. (Source: A European brain for a sovereign website (Dinis Cruz, SGit Newsroom); as read on 20 November 2026.)
- Mistral’s official compressed (Q4_K_M) files are 2.15 GB for the 3B and 5.20 GB for the 8B, plus about 0.85 GB for vision. Those are download sizes, not memory use or speed. (Source: Hugging Face, Ministral-3-3B-Instruct-2512-GGUF and Hugging Face, Ministral-3-8B-Instruct-2512-GGUF; as read on 20 November 2026.)
- Ministral 3 3B, model name
ministral-3b-2512, is listed as Apache 2.0, released 2 December 2025, at 0.10 dollars per million tokens in and 0.10 out. (Source: Mistral docs, Ministral 3 3B and Mistral docs, Pricing; as read on 20 November 2026.) - Mistral’s chat endpoint is
https://api.mistral.ai/v1/chat/completions; the EU one ishttps://api.eu.mistral.ai/v1/chat/completions. (Source: Mistral docs, Chat completion and Mistral docs, Regional inference; as read on 20 November 2026.) - Free mode is the default for new accounts, with the lowest rate limits. Mistral’s key quickstart asks for “a plan with API access”; whether Free mode needs a payment card today was not confirmed. (Source: Mistral docs, Subscriptions and Mistral docs, Activate and generate an API key; as read on 20 November 2026.)
- Ollama lists
ministral-3with the tags3b,8band14b. (Source: Ollama, ministral-3; as read on 20 November 2026.) - Ollama serves an OpenAI-style endpoint at
http://localhost:11434/v1/chat/completions. (Source: Ollama docs, OpenAI compatibility; as read on 20 November 2026.) - OpenRouter’s OpenAI-style API is at
https://openrouter.ai/api/v1. (Source: OpenRouter, Quickstart; as read on 20 November 2026.)
Where the article and Mistral’s pages differ
Where they differ, this site follows Mistral’s own pages:
- EU processing. The article pairs “hosted in the EEA by default” with the EU endpoint. Hosting is not processing: only the EU endpoint commits to processing in the EU. (Source: Mistral docs, Regional inference; as read on 20 November 2026.)
- Training. “Not used for training” holds on Mistral’s inference page, but in Free mode Mistral may train on API data unless you opt out. (Source: Mistral AI, Cloud inference and Mistral help centre, Do you use my data to train and Mistral help centre, Opting out of training; as read on 20 November 2026.)
- Zero data retention. The article says “on paid plans”; Mistral says pay-as-you-go only, on request, and not for Labs or Public Preview models. The article’s link to the zero data retention page (ending
.md) returned 404; the page without.mdworks. (Source: Mistral AI, Privacy policy and Mistral help centre, Zero data retention and Mistral docs, Zero data retention; as read on 20 November 2026.) - OpenAI compatibility. Mistral’s OpenAI-compatible endpoint is
/openai/v1, in public preview;/v1uses its own, similar format. (Source: Mistral docs, Chat completion and Ollama docs, OpenAI compatibility and OpenRouter, Quickstart; as read on 20 November 2026.) - Partnership. The article says there is no public application; Mistral’s partners page has a public expression-of-interest form. (Source: Mistral AI, Partners and Mistral docs, Ambassadors; as read on 20 November 2026.)
- The Large 4 licence. Mistral’s announcement and its help centre describe the condition differently. (Source: Mistral docs, Ministral 3 8B and Mistral AI, Mistral Large 4 and Mistral help centre, Licences of the open models; as read on 20 November 2026.)
- Certifications. Mistral’s wording is that it complies with the frameworks. (Source: Mistral help centre, SOC 2 and ISO 27001; as read on 20 November 2026.)
- The Swarm. The Mistral calls went through OpenRouter, not Mistral’s own API; the model was not recorded. (Source: Pennies against dollars (SGit Newsroom); as read on 20 November 2026.)
- “French company”. Mistral’s About page says European, not French, so this site does not say French. (Source: Mistral AI, About; as read on 20 November 2026.)
Our own calls
This site’s demo agent called Mistral’s API while building the demos on 20 November 2026 (UTC times, from its call log):
- 22:55,
ministral-3b-2512: A five-word hello, to check the answer and its usage fields. 9 prompt tokens, 7 completion, 16 in total. - 23:00,
ministral-3b-2512: The same through the OpenAI-compatible path, /openai/v1. 9 prompt tokens, 6 completion, 15 in total. - 23:03,
ministral-3b-2512: A call from a browser page, to check it can read the answer. 6 prompt tokens, 5 completion, 11 in total. - 23:05,
ministral-3b-2512: The side-by-side demo, end to end (a short brief, at most 150 tokens out). 38 prompt tokens, 150 completion, 188 in total.