French AI startup Mistral announced on Tuesday it will launch its first inference model to compete with rivals such as Openai and China’s Deepseek.
Unlike other inference models, Mistral’s new model brings fresh things to the table that their competitors don’t have. The ability to think about problems in multiple languages.
“We’re publishing a new inference model in a few hours, which has the idiosyncratic nature of being extremely competitive with everyone else and being able to reason in multiple languages,” CEO Arthur Mensch told CNBC in an interview during a fireplace chat at London Tech Week.
Mistral is launching the first inference model to challenge Openai and Deepseek
The launch involves Mistral turning his head and head directly in a serious competition. Openai’s O1 model has been available to ChatGpt users since the end of last year, while Deepseek’s R1 represents an entry into the Chinese reasoning model space. However, Mistral believes there is a gap in the market.
Inference models behave differently than regular AI chatbots. Not only does it spout answers, it also moves complex problems step by step, indicating a logical thinking process. Think of watching someone solve a math problem by writing out each step rather than giving the final answer.
However, Apple’s latest research paper suggests that AI’s inference capabilities may not be able to withstand hype. The company’s researchers have published a paper called “The Illusion of Thinking” that shakes assumptions about these models. They discovered that popular AI systems “facing a complete collapse of accuracy beyond a certain complexity.” Especially when dealing with unfamiliar issues.
However, Mensch says that Mistral’s new model is “good at math.” [and] I’m good at coding. “However, are they real differentiators that go beyond English and Chinese?
“Historically, we have seen the reasons for the English and Chinese models for the reasons,” Mensch explained. Mistral’s future model of therapeutic inference breaks this pattern by supporting European languages.
Why is this important?
This is more important than you think. Most AI development focuses on English, with Chinese models serving that market. European languages have often been added afterwards. For developers and businesses across Europe, having AI that can infer fluently in French, German, Spanish or Italian opens up new possibilities.
Mistral has built a reputation for its large-scale language model of open weight.
These systems expose the underlying parameters and allow developers to peer under the hood to change their core knowledge. This approach saves developers from scratch from the expensive and time-consuming process of training models.
With support from Microsoft, Mistral has the resources to challenge the industry giants. However, the actual test is whether Magistral can offer multilingual promises while consistent with Openai’s O1 performance.
The AI inference model space is rapidly heating up. Each company is trying to solve the same basic problem. This is a way to create an AI system that can not only be pattern matching from training data, but also think through logically complex problems.
Mistral bets that linguistic diversity will be a winning card in this high stakes game.
Founded in 2023 by Arthur Mensch (CEO), Guillaume Lample and Timothy Lacroix, Mistral’s meteor rise reflects a firm commitment to pushing the boundaries of AI innovation. As the company is heading towards full operational status in early 2024, the winds of change continue to drive mistral towards its lofty aspirations in the AI domain.
Mistral Team
🚀Want to share the story?
Submit your stories to TechStartUps.com in front of thousands of founders, investors, PE companies, tech executives, decision makers and tech leaders.
Please attract attention
Source link
