On May 29, 2026, Mistral sent an email to its customers indicating it will retire models on May 31, June 30 and July 31 2026. For each model, there is a newer, better and most of the times, more expensive alternative.

Devstral Small 1.1, Mistral Small 3.2 and Magistral Small 1.2 are replaced by Mistral Small 4 at $0.15/$0.6 per million tokens. For those who were using the first two, it’s a 50% increase on input tokens and 100% increase on output tokens; ouch! For those using Magistral Small 1.2, it’s actually a 70%/60% decrease; nice!

The following models are replaced by Mistral Medium 3.5 at $1.5/$7.5 per million tokens input/output:

  • Mistral Large 2 and Pixtral Large with 25% decrease on input tokens but 25% increase on output tokens
  • Magistral Medium 1.2 with 25% decrease on input tokens but 50% increase on output tokens
  • Devstral Medium 1.1, Devstral 2, Mistral Medium 3 and Mistral Medium 3.1: users will see there a 275% increase on both input and output tokens; outstanding ouch!

Specialized models are retired as well.

  • Mistral OCR 2 is replaced by Mistral OCR 3, the price goes up from $1 to $2 per 1,000 pages. That’s a 100% increase.
  • Voxtral Mini Transcribe 1.0 is replaced by Voxtral Mini Transcribe 2.0. No price indication.
  • Mistral Moderation is replaced by Mistral Moderation 2. No price indication.
  • Leanstral is not replaced. This was a Labs model.

As a user of Devstral 2, I was quite happy with the model quality and price. Unfortunately I can’t say the same about Medium 3.5. My personal experience is that it’s slower, not as “smart” and more rate-limited.

The recent changes after the AI Now Summit 2026 only seem to make it worse. Although they announced having more compute, it does not translate with lower costs, quite the opposite.

I consciously chose Mistral to support an European company, time will tell if I made the wrong choice.