Mistral AI significantly updates its chat AI 'Le Chat' to enable web search functionality and image generation using 'FLUX1.1 [pro]'



Mistral AI, the AI development company behind Mistral Large 2 , which has 123 billion parameters and a 128k context window and recorded benchmark performance comparable to GPT-4o, the multimodal AI Pixtral 12B , and the small-scale model Ministral 3B/8B , has announced a major update to its chat AI, Le Chat.

Mistral has entered the chat | Mistral AI | Frontier AI in your hands
https://mistral.ai/news/mistral-chat/



Mistral AI listed the following six key improvements for 'Le Chat.'

1: Citation-based web search
Le Chat is mostly used by students and professionals for learning, research, work, etc., so to make it easier to use in these cases, the training knowledge has been enhanced with Internet searches. Below is an example of how Mistral AI's marketing team is using AI to evaluate the healthcare market.

Le Chat Web Search - YouTube


2: Canvas for ideas, inline editing and exporting
The canvas is an area displayed separately from the interactive display, and allows you to share output content with Mistral LLM, supporting the creative process of creating something from nothing and realizing ideas. Since the canvas is not bound by two-way dialogue, you can use the powerful inference capabilities of LLM to directly edit and transform content. Below, the Mistral AI product team is creating a mockup of a product metrics dashboard in React in preparation for the launch of the canvas.

Canvas - YouTube


3. Understanding documents and images using the new multimodal AI 'Pixtral Large'
With the power of 'Pixtral Large', it is now possible to read PDF files attached to the input area and process documents and images. Below, I am loading the 1935 paper that was the basis of the ' Einstein-Podolsky-Rosen Paradox ' and extracting and summarizing information.

Research Paper OCR - YouTube


4: Image generation using ' FLUX1.1 [pro] '
High-quality image generation can be achieved using Black Forest Labs' image generation AI 'FLUX1.1 [pro]'. Examples of work using FLUX1.1 [pro] can be seen in the following article.

Black Forest Labs, an AI company created by the original developers of Stable Diffusion, releases image generation AI FLUX1.1 [pro] and BFL API for application integration - GIGAZINE



5. Fully integrated from model to output

6: Rapid response with speculative editing

A comparison with major chat AIs and search assistants is as follows:

Functions \ Chat AI ChatGPT Perplexity Claude Le Chat
Access to the latest Frontier models ○ (Charged)
Cited Web Search
Canvas Function ○ (Charged) ○ (Limited)
Image Generation
Image and document analysis ○ (Charged)
Task Agent ○ (Charged)
A single platform from model to output?



The beta version of Le Chat is available for free. Mistral AI says, 'Our mission is not to aim for AGI (artificial general intelligence) at all costs, but to put frontier AI in your hands and let you decide for yourself what to do with advanced AI capabilities.'

in Software,   Video, Posted by logc_nt