Devstral Models by MistralAI
3 models2025Up to 256k ctxFrom $0.1/1M input
Capabilities
Vision1 of 3 models
Multimodal1 of 3 models
Function Calling2 of 3 models
Tool Use2 of 3 models
Structured Outputs2 of 3 models
Code Execution2 of 3 models
About
Devstral is a family of 3 AI models by MistralAI, released in 2025.
Current Variants
Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.
3 in view
Devstral Small 2Current
Use when the workload needs code, 256k context, and 24B parameters.
2025-12code256k context24B parameters
Mistral Devstral 2 123BCurrent
Use when the workload needs code, 123B parameters, and structured outputs.
2025-12code123B parametersstructured outputs
Devstral Medium 2507Current
Use when the workload needs code, 131k context, and tool use.
2025-07code131k contexttool use
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Devstral Small 2 | Use when the workload needs code, 256k context, and 24B parameters. | 2025-12 | code256k context24B parameters | Current |
| Mistral Devstral 2 123B | Use when the workload needs code, 123B parameters, and structured outputs. | 2025-12 | code123B parametersstructured outputs | Current |
| Devstral Medium 2507 | Use when the workload needs code, 131k context, and tool use. | 2025-07 | code131k contexttool use | Current |
Release Timeline
2 release groups2025-12
2 current
Devstral Small 2
Currentcode256k context24B parameters
Mistral Devstral 2 123B
Currentcode123B parametersstructured outputs
2025-07
1 current
Devstral Medium 2507
Currentcode131k contexttool use
Specifications(3 models)
| Model | Released | Context | Parameters | Vision | Multimodal | Fn Calling | Tool Use | Structured Outputs | Code Exec |
|---|---|---|---|---|---|---|---|---|---|
| Devstral Small 2 | 2025-12 | 256k | 24B | Yes | Yes | Yes | Yes | Yes | Yes |
| Mistral Devstral 2 123B | 2025-12 | — | 123B | No | No | No | No | Yes | Yes |
| Devstral Medium 2507 | 2025-07 | 131k | — | No | No | Yes | Yes | No | No |
Available From(3 providers)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| Devstral Small 2 | Vercel AI Gateway | $0.1 | $0.3 | Serverless |
| Devstral Small 2 | Mistral AI Studio | $0.1 | $0.3 | Serverless |
| Mistral Devstral 2 123B | AWS Bedrock | $0.4 | $2 | Serverless |
| Mistral Devstral 2 123B | Mistral AI Studio | $0.4 | $2 | Serverless |
Frequently Asked Questions
- What is Devstral used for?
- Devstral is used for code, vision and multimodal work, and agent workflows and tool use. The family description and listed model capabilities point to those workloads as the best fit.
- How does Devstral compare to Ministral?
- Devstral by MistralAI is strongest where you need code, while Ministral by MistralAI is the closest related family to check for vision and multimodal work. Devstral has 3 listed variants and reaches up to 256k context, while Ministral reaches up to 32k context, so compare the specs and pricing tables before choosing a production model.
- Which Devstral model should I use?
- For the lowest listed input price, start with Devstral Small 2 through Mistral AI Studio at $0.1/1M input tokens. For the most capable/latest local choice, evaluate Devstral Small 2 with 256k context and tool use, function calling, structured outputs, and multimodal inputs.
