This analysis describes what Mistral AI's agreement states, permits, or reserves. It does not constitute a legal determination about enforceability. Regulatory applicability and practical outcomes may vary by jurisdiction, enforcement context, and individual circumstances. Read our methodology
A 256k-token context ceiling directly bounds how much input and conversational history the model can consider in a single request.
Interpretive note: The excerpt is a two-word label and value with no surrounding explanation. '256k' is interpreted as 256,000 tokens, which is the standard industry reading, but this is not confirmed by surrounding text.
The updated model card reflects new pricing for mistral-medium-3-5: the token cost has increased from €1.25 per million tokens (€6.4 per million on a higher tier) to $1.5 per million tokens ($7.5 on a higher tier). The pricing currency has shifted from EUR to USD as the primary denomination. Organizations using this model through Mistral AI's API will encounter higher per-token costs at current exchange rates. No opt-out mechanism is available; the new pricing applies to all API consumption of this model.
View change record →Any single request to the model cannot exceed 256,000 tokens of combined input context.
How other platforms handle this
Certain aspects and features of our Services and Sites are only available through the use of Cookies, so if you choose to disable or decline Cookies through your device or browser settings, your use of the Sites and Services may be limited...
These choices do not apply to certain informational communications including important information and documentation relating to your account.
We've distilled our commitment into six core values: Fairness Reliability and safety Privacy and security Transparency Accountability Inclusiveness
"Context 256kExcerpt from Mistral AI's Mistral Medium 3.5 Model Card
Get the research letter
Companies change their terms quietly. We read every version and catch what actually changed. One email a week on the changes that matter and what they mean.
A 256k-token context ceiling directly bounds how much input and conversational history the model can consider in a single request.
Any single request to the model cannot exceed 256,000 tokens of combined input context.
No. ConductAtlas is an independent monitoring service. We are not affiliated with, endorsed by, or sponsored by Mistral AI.