Mistral Moves Coding Agents to the Cloud with Mistral Medium 3.5
Mistral
Mistral AI has released Mistral Medium 3.5, a 128B open-weight model optimized for long-horizon coding and productivity. The update introduces remote agents in Mistral Vibe and Le Chat, enabling parallel asynchronous task execution, along with a new Work mode for complex multi-step tasks. The model is available today on Pro, Team, and Enterprise plans, priced at $1.5 per million input tokens and $7.5 per million output tokens.
Mistral AI has moved coding agents to the cloud, allowing parallel asynchronous task execution. The new Mistral Medium 3.5 model is a 128B dense model with a 256k context window, released as open weights under a modified MIT license. It scores 77.6% on SWE-Bench Verified and 91.4 on τ³-Telecom. Remote agents can be started from the Mistral Vibe CLI or Le Chat, and sessions run in isolated sandboxes. Work mode in Le Chat enables complex multi-step tasks, with human oversight for sensitive actions. The model is available on Pro, Team, and Enterprise plans, and also via API at $1.5 per million input tokens and $7.5 per million output tokens, on NVIDIA GPU-accelerated endpoints and as a NIM microservice.
- Сокращения
- API = Application Programming Interface — интерфейс программирования приложений
- CLI = Command Line Interface — интерфейс командной строки
- GPU = Graphics Processing Unit — графический процессор
- PR = Pull Request — запрос на слияние
- NIM = NVIDIA Inference Microservice — микросервис инференса NVIDIA
- MIT = Massachusetts Institute of Technology — лицензия MIT
Source: Mistral AI —
original