Model comparison
Codestral 2508 vs Devstral 2 2512
Compare Codestral 2508 and Devstral 2 2512: input/output $/Mtoken, context window, modalities, and license. OpenRouter-synced pricing.
Mistral
Codestral 2508
Mistral's cutting-edge language model for coding released end of July 2025. Codestral specializes in low-latency, high-frequency tasks such as fill-in-the-middle (FIM), code correc
Mistral
Devstral 2 2512
Devstral 2 is a state-of-the-art open-source model by Mistral AI specializing in agentic coding. It is a 123B-parameter dense transformer model supporting a 256K context window. De
| Metric | Codestral 2508 | Devstral 2 2512 |
|---|---|---|
| Provider | Mistral | Mistral |
| Context window | 256,000 | 262,144 |
| Input $/Mtok | $0.300 | $0.400 |
| Output $/Mtok | $0.900 | $2.00 |
| Modalities | text | text |
| License | open-weights | open-weights |
Quick take
On input price, Codestral 2508 is cheaper at $0.300/Mtok. For context window, Devstral 2 2512 leads with 262,144 tokens.
Pick based on your workload: high-volume cheap inference vs long-document / agent loops. Enterprise DNA can wire either model into Omni with evals, secrets, and job orchestration.