The Browser Company launches new AI browser Dia
1031
English
English
French artificial intelligence startup Mistral AI has released its latest open source model, Mistral Small 3.1. This "mini" model with only 24 billion parameters can actually match or even surpass similar products from American technology giants such as Google and OpenAI, injecting new competitive vitality into the AI market.
Mistral Small 3.1 has significantly improved text performance , has multimodal comprehension capabilities , and is able to process text and images. Its context window extends to 128k tokens , allowing longer information sequences to be processed at once. In addition, the model's information processing speed is as high as 150 tokens per second , making it ideal for applications with high response speed requirements.
Mistral AI chose to release Mistral Small 3.1 under a loose Apache 2.0 license , demonstrating the differences between closed proprietary systems and open alternatives in the AI industry. Through the open source strategy, Mistral hopes to attract global developers to innovate based on their models and expand R&D capabilities.
Founded in 2023 by former researchers at Google DeepMind and Meta, Mistral AI is currently valued at about $6 billion, making it a leading AI startup in Europe. Its chat assistant Le Chat received 1 million downloads within two weeks of the release of the mobile version, and French President Macron even called on the public to use Le Chat instead of OpenAI's ChatGPT.
Mistral AI is positioned as the " world's greenest and leading independent AI lab ", emphasizing European digital sovereignty to distinguish it from its US competitors. With the entry into force of the European Union's Artificial Intelligence Act, European companies like Mistral that focus on compliance have more advantages.
The highlight of Mistral Small 3.1 is its amazing efficiency . In the context of the general pursuit of larger models and massive computing resources in the AI industry, Mistral focuses on algorithm improvement and training optimization to tap the greatest potential in smaller architectures. This approach effectively addresses the key challenges facing AI deployment: huge computing and energy costs. Mistral's model can even run on an RTX4090 graphics card or a Mac with 32GB of memory, allowing advanced AI technology to be applied to a wider range of scenarios, including device-side applications.
The release of Mistral Small 3.1 shows a more efficient, open and European-style AI development path . This rising European AI star may really shake the dominance of American technology giants in this field.
Model: https://top.aibase.com/tool/mistral-small-3-1
Official blog: https://mistral.ai/news/mistral-small-3-1
