|
|
|
|
August 21, 2025
|
Hackers Infiltrate Alleged North Korean Operative’s Computer, Leak Evidence of...
|
|
August 21, 2025
|
Ecosia Proposes Unusual Stewardship Model for Google Chrome
|
|
August 21, 2025
|
OpenAI Presses Meta for Evidence on Musk’s $97 Billion Takeover Bid
|
|
August 15, 2025
|
ChatGPT Mobile App Surpasses $2 Billion in Consumer Spending, Dominating Rivals
|
|
|
French Startup Mistral Launches Voxtral, an Open Audio AI Model Challenging Corporate Giants
July 15, 2025
As AI systems become increasingly advanced, speech is rapidly turning into the default mode of communication with machines. French AI startup Mistral has entered the audio AI race with the release of its first open model, aiming to disrupt the market dominated by closed corporate systems by offering open-weight alternatives.
On Tuesday, Mistral unveiled Voxtral, its inaugural family of audio models designed specifically for business use.
The company markets Voxtral as the first truly open model capable of delivering “usable speech intelligence in production.” This means developers no longer have to choose between cheaper open systems that often misinterpret speech and closed, high-quality models that come with hefty price tags and limited deployment control.
For businesses, Voxtral promises an affordable alternative, with Mistral claiming it costs “less than half the price” of comparable solutions on the market.
Mistral explains that Voxtral can transcribe audio clips up to 30 minutes in length. Thanks to its large language model backbone, Mistral Small 3.1, it can process and understand up to 40 minutes of audio, enabling users to ask questions about the content, generate summaries, or convert voice commands into real-time actions such as calling APIs or executing functions. Moreover, Voxtral supports multiple languages, including English, Spanish, French, Portuguese, Hindi, German, Dutch, and Italian.
The company offers two versions of its “speech understanding models.” The first, Voxtral Small, has 24 billion parameters suited for production-scale deployments and is competitive with solutions like ElevenLabs Scribe, GPT-4o-mini, and Gemini 2.5 Flash.
The second variant, Voxtral Mini, is a lighter model with 3 billion parameters optimized for local and edge deployments. Alongside this, Mistral offers Voxtral Mini Transcribe, a stripped-down, ultra-affordable API version designed specifically for transcription tasks. This version aims to outperform OpenAI’s Whisper while costing less than half as much.
Users interested in testing Voxtral can access the API for free on Hugging Face or try out the models in Mistral’s chatbot, Le Chat. Integration pricing starts at $0.001 per minute, making it an attractive option for developers and companies alike.
This launch follows last month’s announcement of Magistral, Mistral’s first family of reasoning models that improve reliability by solving problems step-by-step.
Mistral, widely recognized as one of Europe’s leading AI firms, has built a reputation for advocating open-source AI models. Recently, TechCrunch reported that Mistral is in talks to raise up to $1 billion in equity funding from investors including Abu Dhabi’s MGX fund, further fueling its ambitions to challenge major industry players with innovative, accessible AI technologies.
|
|
|
Sign Up to
Our Newsletter!
Get the latest news in tech.
|
|
|
