Mistral: Devstral 2 2512

Pricing, context window and lifecycle for mistralai/devstral-2512, tracked from what the vendor publishes.

2026-07-31 - its vendor retired this model on that date, and the catalogue still lists it. Anything still calling it is running on borrowed time.

To find every place that calls it:

curl -sO https://neosignal-ai.vercel.app/check.py
python check.py .

Price per million tokens

Input$0.400
Output$2.00
Cache price (read)$0.040
Max price$2.00

Facts

Model IDmistralai/devstral-2512
VendorMistral
Token limit (context window)262,144 tokens
Knowledge cutoff-
Retired on2026-07-31
Date fromvendor deprecation page

Supported parameters

frequency_penalty, max_tokens, presence_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools, top_p

Other Mistral models

Change history

Every change we have observed since tracking began.

DetectedChangeDetail
2026-08-30 19:14Pricing ChangeOutput $2.20 -> $2.00 (-9.1%) · Cache price (read) $0.044 -> $0.040 (-9.1%) · Input $0.440 -> $0.400 (-9.1%)
2026-08-24 19:14Model AddedMistral: Devstral 2 2512
2026-08-01 01:02Model RemovedMistral: Devstral 2 2512

Check this model and the rest of your stack It opens with this id already filled in. Nothing is stored - the list lives in the link.

Subscribe to this model by RSS - just this one, no signup.