ColinBuilds · Retired on Mistral API · last listed price kept
Devstral 2
- Input / 1M
- $0.40
- Output / 1M
- $2.00
- Mistral-published context window
- 256,000 tokens
Last checked Mistral API pricing for Devstral 2 / devstral-medium-latest was $0.40 input and $2.00 output per 1M tokens. Mistral's models overview lists Devstral 2 (devstral-2512) as deprecated 2026-05-22 and retired 2026-07-31, with Mistral Medium 3.5 as the alternative. Calculator keeps the last listed price for historical comparison. · as of August 25, 2026
Compare models
Same per-1M input/output rates as above. Enter your monthly token volumes to compare bills.
Devstral 2
- Input: 1M × $0.40 = $0.40
- Output: 1M × $2.00 = $2.00
- Total
- $2.40
GPT-4o
- Input: 1M × $2.50 = $2.50
- Output: 1M × $10.00 = $10.00
- Total
- $12.50
Difference
$10.10
GPT-4o costs $10.10 more than Devstral 2.
Popular comparisons
37 total featuring Devstral 2.
Devstral 2 was Mistral’s Devstral coding model for software-engineering agents. Mistral’s current models overview lists it as retired. The last listed API price is kept here for historical comparison.
What this model is
Mistral’s models overview lists Devstral 2 v25.12 with API ID devstral-2512 under Deprecated, with deprecation on 2026-05-22 and retirement on 2026-07-31. The listed alternative is Mistral Medium 3.5.
The earlier pricing page described Devstral 2 as an open-weights agentic coding model and showed the live pricing row as devstral-medium-latest. That live row is no longer present on the current Mistral API pricing page checked on 2026-08-25.
Pricing notes
The calculator on this page keeps the last listed official Mistral API rates for Devstral 2 / devstral-medium-latest: $0.40 per 1M input tokens and $2.00 per 1M output tokens.
Treat those figures as historical last-listed prices, not as a current purchasable Mistral API route. For current Mistral coding and agent work, Mistral points to Mistral Medium 3.5.
Benchmarks and specs
Mistral’s model card listed Devstral 2 v25.12 with the API ID devstral-2512, an OPEN badge, and a 256k-token context window.
No benchmark score is shown until an exact benchmark source and model identity are matched to Devstral 2.
Best fit
Devstral 2 is kept for historical Mistral coding-agent cost comparisons. It should not be used as a current default Mistral API recommendation.
Compare current work against Mistral Medium 3.5, Mistral Large 3, Mistral Small 4, or Codestral.