The French-made AI model 'Mistral Medium 3.5,' which surpasses Claude Sonnet 4.5, has been released as an open model, and Mistral Vibe now supports cloud computing.

French AI company Mistral AI has released its AI model ' Mistral Medium 3.5, ' with 128 billion parameters, as an open model. In addition, its coding agent, Mistral Vibe, now runs in the cloud, allowing multiple agents to run simultaneously and tasks to be initiated from the chat AI, Le Chat.
Remote agents in Vibe. Powered by Mistral Medium 3.5. | Mistral AI
Mistral Medium 3.5, a new flagship model in public preview by @MistralAI that merges instruction-following, reasoning, and coding into a single 128B dense model with a 256k context window and configurable reasoning effort. It's a new default model for Mistral Vibe and Le Chat.… pic.twitter.com/vcW1GaB3s1
— Mistral Vibe (@mistralvibe) April 29, 2026
Mistral Medium 3.5 is an AI model with 128 billion parameters and a context length of 256,000. It boasts high coding agent capabilities and supports dozens of languages, including English, Japanese, and French. It can also accept image input in addition to text.
The graph below compares the agent performance of 'Mistral Medium 3.5,' 'Devstral 2,' and 'Devstral 2 Small.' It clearly shows that the performance of Mistral Medium 3.5 has significantly improved compared to the previous model.

The agent performance comparison results for 'Mistral Medium 3.5 (128B)', 'Claude Sonnet 4.5', 'Claude Sonnet 4.6', 'Kimi K2.5 (1T A32B)', 'GLM-5.1 (744B A40B)', and 'Qwen3.5 (397B A17B)' are as follows. Mistral Medium 3.5 outperformed Claude Sonnet 4.5 in most tests.

In tests measuring mathematical performance, it outperforms larger Chinese-made models. Mistral AI promotes the Mistral Medium 3.5 as 'a system that can run on just four GPUs, yet delivers outstanding real-world performance.'

Mistral Medium 3.5 is positioned as a public preview version and is being deployed as the default model for Mistral Vibe and Le Chat. The model is also available on Hugging Face, and Mistral Medium 3.5 128B EAGLE, which supports speculative inference, has also been released.
mistralai/Mistral-Medium-3.5-128B · Hugging Face
https://huggingface.co/mistralai/Mistral-Medium-3.5-128B
mistralai/Mistral-Medium-3.5-128B-EAGLE · Hugging Face
https://huggingface.co/mistralai/Mistral-Medium-3.5-128B-EAGLE
In addition, Mistral Vibe now runs in the cloud, allowing agent tasks that were being processed locally to be handed over to the cloud. The integration between Mistral Vibe and Le Chat has also been enhanced, allowing agent tasks to be initiated from conversations with Le Chat.
Teleport your coding session to the cloud - YouTube
Related Posts:







