SIDE BY SIDE
Codestral 2508 vs Devstral 2 2512
Both are Codestral & Devstral models. Devstral 2 2512 was added to OpenRouter on 2025-12-09, Codestral 2508 on 2025-08-01.
| Field | Codestral 2508 | Devstral 2 2512 |
|---|---|---|
| Provider | Mistral | Mistral |
| Family | Codestral & Devstral | Codestral & Devstral |
| Province | The Cedar Forest | The Cedar Forest |
| Context window | 256K tokens | 262K tokens |
| Maximum output | 205K tokens | 210K tokens |
| Input / 1M tokens | $0.300 | $0.400 |
| Output / 1M tokens | $0.900 | $2.00 |
| Accepts | text, files | text, files |
| Produces | text | text |
| Reasoning | No | No |
| Tool use | Yes | Yes |
| Added to OpenRouter | 2025-08-01 | 2025-12-09 |
| Model ID | mistralai/codestral-2508 | mistralai/devstral-2512 |
Questions and answers
Which is cheaper, Codestral 2508 or Devstral 2 2512?
Codestral 2508 costs less for both input ($0.300 vs $0.400) and output ($0.900 vs $2.00) per million tokens. On input, Codestral 2508 is 25% cheaper.
Which has the larger context window?
Devstral 2 2512: 262K tokens against 256K for Codestral 2508 (1× as long).
What can one do that the other cannot?
Nothing among reasoning, tool use and image input: both support tool use.
Answers are computed from the 2026-10-24 catalog snapshot.