Le Chat Fully Competes with ChatGPT: Europe's Rising AI Star Strives to Catch Up}

Mistral AI is just trying to make a European version of OpenAI?

In recent months, the European AI startup Mistral AI, founded by former Google and Meta researchers, has been making waves with a series of open-source models spanning various domains, including the "world’s best" OCR model, a multimodal model comparable to Claude, the first reasoning large model Magistral, and the recently released Voxtral, a top-tier open-source speech model.

It seems this European AI newcomer is not satisfied yet—they want to compete fiercely with OpenAI at the application level. They have upgraded Le Chat once again, adding powerful new features that make it more robust, intuitive, and fun, almost fully rivaling ChatGPT in capabilities.

Le Chat’s New Features

• Deep Research Mode: Quickly generate structured research reports on complex topics.
• Voice Mode: Interact with Le Chat using our new Voxtral model, no keyboard needed.
• Native Multilingual Reasoning: Use Magistral, our reasoning model, to get thoughtful answers.
• Project Management: Organize your conversations into content-rich folders.
• Advanced Image Editing: Directly edit images within Le Chat, in collaboration with Black Forest Labs.

      ^{Le Chat’s Promotional Video}

Research Mode transforms Le Chat into a coordinated research assistant capable of planning, clarifying needs, searching, and synthesizing information. It can break down complex questions, gather reliable data, and generate well-structured, reference-supported reports.

Driven by tool-enhanced deep research agents, designed to be simple, transparent, and genuinely helpful, it feels like working with a well-organized research partner.

Mistral AI also showcases use cases on its official website. The research mode can track market trends, write business strategies, create personal plans, and most importantly, conduct academic research.

Voice Mode: Chat naturally with Le Chat—no typing needed. Use our new Voxtral speech input model, designed for natural, low-latency speech recognition that keeps up with your workflow.

Currently, Le Chat only supports speech-to-text input, not real-time voice conversations, so the digital companion experience like Grok 4 Ani is still out of reach.

In image editing, users can create and modify images with simple prompts like “remove objects” or “place me in another city.” The model supports scene transformation while maintaining character and detail consistency, ensuring edits are seamless and coherent across images.

Many users have shared positive experiences on forums, praising Le Chat’s image editing capabilities as superior to OpenAI’s. One user remarked, “OpenAI’s models tend to alter the entire image during editing, causing errors in unrelated areas. Le Chat seems to preserve unaffected parts perfectly and applies edits selectively, which is impressive!”

One user uploaded a photo of a home office and requested: “Fix the slightly torn gray panels at the bottom to look brand new.” The result was highly satisfactory.

Another example shows the original image and the edited version, demonstrating Le Chat’s advanced editing capabilities.

These new features are well-known to our readers. After this major update, Le Chat now closely matches the capabilities of industry-leading products like ChatGPT.

Recently, Mistral AI’s rapid progress has demonstrated Europe’s strong momentum in large models. Many users expressed excitement about Mistral’s swift advancements.

Interestingly, the name “Mistral” is also associated with a pixel cat at the bottom of their homepage, and their logo resembles a cute cat head, adding a playful touch to their branding.

Mistral AI’s Speech Recognition Model

On July 15, Mistral AI released Voxtral, a new speech recognition model claimed to be the “world’s best (and open source).”

Voxtral surpasses Whisper large-v3 in speech transcription, outperforming GPT-4o mini Transcribe and Gemini 2.5 Flash across all tasks, with excellent results in English and other major languages, demonstrating strong multilingual capabilities.

Voxtral 3B and Voxtral 24B models not only support speech transcription but also feature:

• Extended Context Understanding: Supports up to 32k tokens, transcribes audio up to 30 minutes long, and understands content for up to 40 minutes.
• Built-in Q&A and Summarization: Directly ask questions or generate structured summaries from audio without needing to connect speech recognition with language models.
• Multilingual Support: Automatic language detection with top performance across major languages like English, Spanish, French, Portuguese, Hindi, German, Dutch, and Italian, serving global users with a single system.
• Voice-Triggered Function Calls: Trigger backend functions, workflows, or APIs directly through voice commands, enabling seamless voice-to-system instructions.
• Powerful Text Understanding: Continues to leverage high-performance language models (based on Mistral Small 3.1) for advanced text processing.