Head to head
GPT-5 mini vs Devstral 2
On a blended workload of one million input tokens and three million output tokens, GPT-5 mini is 1.0× cheaper than Devstral 2.
| GPT-5 mini | Devstral 2 | |
|---|---|---|
| Provider | OpenAI | Mistral |
| Input / MTok | $0.25 | $0.4 |
| Output / MTok | $2 | $2 |
| Cache read / MTok | $0.025 | — |
| Context window | — | — |
| Blended 1M in + 3M out | $6.25 | $6.40 |
| Status | ga | ga |