Model Library
MistralReleased December 9, 2025Devstral 2
Mistral · Released December 9, 2025
Intelligence score
19
At a glance
- Intelligence
- 19
- Speed
- 76.3tokens/sec
- Latency
- 1.22stime to first token
- Input price
- Not published
- Output price
- Not published
- Context window
- 260Ktokens
- Parameters
- 125billions
Summary
Comparison summary
Devstral 2 scores 19 on the Artificial Analysis Intelligence Index, placing it well above average among other open weight non-reasoning models of similar size (median: 7). Devstral 2 generates output at 76.3 tokens per second (based on Mistral's API), which is below average compared to other open weight non-reasoning models of similar size (median: 85.6 t/s).
Specifications
Technical specifications
| Model name | Devstral 2 |
|---|---|
| Provider | Mistral |
| Release date | December 9, 2025 |
| Intelligence score | 19 |
| Speed | 76.3 tokens/sec |
| Latency (TTFT) | 1.22s |
| Input price | Not published |
| Output price | Not published |
| Context window | 260,000 tokens |
| Parameters | 125B |
| Reasoning | No |
| Input modalities | text |
| Output modalities | — |
| Open weights | Yes |
| License | Modified MIT License |
| API availability | — |
| Source | Model Intelligence legacy profile: devstral-2 |
FAQ · 18
Frequently asked questions
Devstral 2 was released on December 9, 2025.
Devstral 2 was created by Mistral.
Devstral 2 scores 19 on the Artificial Analysis Intelligence Index, placing it well above average among other open weight non-reasoning models of similar size (median: 7).
Devstral 2 generates output at 76.3 tokens per second (based on Mistral's API), which is below average compared to other open weight non-reasoning models of similar size (median: 85.6 t/s).
Devstral 2 has a time to first token (TTFT) of 1.22s (based on Mistral's API), which is better than average compared to other open weight non-reasoning models of similar size (median: 1.59s).
When evaluated on the Intelligence Index, Devstral 2 generated 9.2M output tokens, which is better than average compared to other open weight non-reasoning models of similar size (median: 9.2M).
No, Devstral 2 is not a reasoning model. It provides direct responses without extended chain-of-thought reasoning.
Devstral 2 supports text input.
Devstral 2 supports text output.
No, Devstral 2 does not support image input. It can only process text.
No, Devstral 2 is not multimodal. It only supports text input.
Devstral 2 has a context window of 260k tokens. This determines how much text and conversation history the model can process in a single request.
Yes, Devstral 2 is open weights. The model weights are publicly available and can be downloaded for self-hosting.
Devstral 2 has 125 billion parameters.
Devstral 2 is released under the Modified MIT License license. Commercial use requires a separate license agreement.
Devstral 2 achieves a score of 19 on the Artificial Analysis Intelligence Index. This composite benchmark evaluates models across reasoning, knowledge, mathematics, and coding.
Yes, Devstral 2 is available via API through 1 provider.
Devstral 2 is available through 1 API provider.