Mistral’s New Ultra-Fast Translation Model Gives Big AI Labs a Run for Their Money

Mistral AI has released a new family of AI models that it claims will clear the way to seamless conversation between people who speak different languages.

On Wednesday, the Paris-based AI lab released two new speech-to-text models: Voxtral Mini Transcribe V2 and Voxtral Realtime. The former is built to transcribe audio files in large batches and the latter for near real-time transcription, within 200 milliseconds; both can translate between 13 languages. Voxtral Realtime is freely available under an open source license.

At four billion parameters, the models are small enough to run locally on a phone or laptop — a first in the speech-to-text field, Mistral claims — meaning private conversations don't need to be sent to the cloud. According to Mistral, the new models are both cheaper to run and less error-prone than competing alternatives.

Mistral has Voxtral Realtime – although the model outputs text, not speech – as a marked step towards free-flowing conversation over the language barrier, a problem Apple and Google are also competing to solve. Google's latest model is capable translate with a delay of two seconds.

“What we are building is a system to be able to translate seamlessly. This model basically lays the groundwork for that,” says Pierre Stock, VP of Science Operations at Mistral, in an interview with WIRED. “I think this problem will be solved in 2026.”

Founded in 2023 by Meta and Google DeepMind alumni, Mistral is one of few European companies developing fundamental AI models capable of running remotely close to the American market leaders – OpenAI, Anthropic, and Google – from a capacity point of view.

Without access to the same level of funding and computing, Mistral has focused on performance through imaginative model design and careful optimization of training datasets. The goal is that micro-improvements across all aspects of model development translate into material performance gains. “Honestly, too many GPUs make you lazy,” Stock claims. “You just test a lot of things blindly, but you don't think what is the shortest path to success.”

Mistral's flagship large language model (LLM) does not match competing models developed by American competitors for raw assets. But the company has carved out a brand by striking a compromise between price and performance. “Mistral offers an alternative that is more cost efficient, where the models are not as big, but they are good enough, and they can be shared openly,” says Annabelle Gawer, director at the Center of Digital Economy at the University of Surrey. “It may not be a Formula 1 car, but it is a very efficient family car.”

Meanwhile, as its American colleagues throw hundreds of billions of dollars at the race for artificial general intelligence, Mistral is building a list of specialist – albeit less sexy – models intended to perform narrow tasks, such as converting speech into text.

“Mistral does not position itself as a niche player, but it certainly makes specialized models,” says Gawer. “As an American player with resources, you want to have a very powerful technology for general purposes. You don't want to waste your resources by fine-tuning it to the languages ​​and specificities of certain sectors or geographies. You leave this kind of less profitable business on the table, which creates space for middle players.”

As the relationship between the US and its European allies shows signs of deterioration, Mistral has also been increasingly drawn to its European roots. “There is a trend in Europe where companies and in particular governments are looking very carefully at their dependence on American software and AI companies,” says Dan Bieler, principal analyst at IT consulting firm PAC.

Against that backdrop, Mistral has positioned itself as the safest pair of hands: a European-native, multilingual, open source alternative to the proprietary models developed in the US. “Their question has always been: How do we build a defensible position in a market that is dominated by hugely funded American actors?” says Raphaëlle D'Ornano, founder of tech consultancy D'Ornano + Co. “The approach that Mistral has taken so far is that they want to be the sovereign alternative, in accordance with all the regulations that may exist within the EU.”

Although the performance gap for the US heavyweights will remain, as companies struggle with the need to find a return on AI investment and factor in the geopolitical context, smaller models tailored to industry and region-specific requirements will have their day, Bieler predicts.

“The LLMs are the giants that dominate the discussions, but I wouldn't count on this being the situation forever,” Bieler claims. “Small and more regionally focused models will play a much bigger role in the future.”