The business was valued at US$2 billion in 180 days! The European version of OpenAI explodes in popularity, Llama's creators set up their own business, and NVIDIA has become a shareholder

The latest valuation of "European version of OpenAI" is approaching US$2 billion!

Mistral AI, a large model startup based in Paris, latest round of financing, $487 million.

Investors include Andreessen Horowitz, Nvidia, Salesforce, etc.

Among them, Andreessen Horowitz led the investment of approximately 200 million euros, and Nvidia and Salesforce plan to invest another 120 million euros through converted bonds.

It is understood that Mistral AI was founded by former DeepMind and former Meta scientists. The two main creators are the core contributors of Llama.

In May this year, they started their own business after running away, which quickly attracted attention in the industry. The seed round received $113 million, and the company was valued at $260 million.

In September, the open source large model Mistral-7B was released, fully benchmarking Llama2.

Now, as the latest round of financing enters the final stage, it means that Mistral AI

only needs 6 months to successfully become a unicorn.

According to the introduction, these new funds will be used to expand the size of the team and accelerate the research and development process

The main creator of Llama left to found

Mistral AI was established in May this year and is headquartered in Paris , the company’s mission is “to build next-generation artificial intelligence systems that can solve real-world problems.”

Because it has a grand AI vision, its main creators are all top talents in the AI ​​field, and it quickly releases self-developed large models, Mistral AI is regarded by the industry as the "European version of OpenAI".

The three founders of the company are:

CEO: Arthur Mensch
• Chief Scientist: Guillaume Lamper (Guillaume Lample)
• CTO: Timothée Lacroix

The chief scientist and CTO are both the core creators of

Llama .

Chief scientist Guillaume Lamper was a research scientist at Meta AI and led the development of Llama.

He received his PhD from Carnegie Mellon University, studying under Chris Dyer, focusing on natural language processing (NLP), machine learning and other research fields

Google Scholar shows that Rampur’s paper has been cited 19,000 times.

Timothy Lacroix is ​​a CTO who has worked at Meta AI for 8 years and is also a major contributor to PyTorch-BigGraph

CEO Arthur Menchi is 30 years old this year. He worked as a researcher at DeepMind’s Paris branch before resigning to start a business in 2023.

While working at DeepMind, Menchi became The main contributors to projects such as Retro, Flamingo, Chinchilla, etc., have accumulated rich experience in the development and optimization of large language models

#It is understood that the three main creators met in school, when they They are all studying in the field of AI. CEO Menqi said that they began to discuss the idea of ​​​​AI entrepreneurship in 2022.

In June of this year, just four weeks after its establishment, Mistral AI won a $113 million seed round

, led by top venture capital Lightspeed, with a valuation of $260 million. At that time, the team stated that they would be committed to

large model open source work

. In September, Mistral AI officially released and open sourced its first large model

Mistral-7B

, and fully benchmarked it against Llama2. Mistral-7B surpasses Llama 2 (13B) in all benchmarks and surpasses Llama 2 (34B) in reasoning, mathematics, and code generation tasks, according to the paper

Mistral-7B adopts the same group query attention mechanism in Llama2, which can accelerate reasoning; combined with the sliding window attention mechanism, it can effectively handle sequences of any length and reduce reasoning costs.

At the same time, they also provide a fine-tuned model

Mistral 7B – Instruct

, which also surpasses the performance of the Llama2 (13B) chat model in human and automated benchmarks.

Mistral-7B shows that LLM’s knowledge compression capabilities may exceed our previous expectations. If you want to get better performance with small models, there is still a lot of room for exploration.

According to our understanding, in order to avoid copyright issues, Mistral chose to use public data to train large-scale models

Become a rising European AI star

With the release of its own large model, Mistral officially entered the In the competition with OpenAI, Google, etc.

CEO Menqi once said in an interview that although the company is very young and the team is small, "we compete with everyone".

Due to the expertise of a large number of top talents in the field of large models, Mistral has attracted much attention in the industry since its establishment, especially in Europe

Its seed round attracted many well-established European venture capitals, such as France’s Redpoint and Index Ventures, Germany’s La Famiglia and Headline, as well as Italian, Belgian, and British venture capitals also participated.

At the first Artificial Intelligence Security Summit held in London not long ago, Mistral, as the only European company, attended the second day's small meeting with technology giants.

Moreover, Mistral did not disappoint investors.

It is said that the release of its first large-scale model was originally expected to be released at the end of the year. As a result, it was released 3 months early in September.

Having a top team and a large self-developed model explains why Mistral’s valuation has grown rapidly

This latest round of financing was mainly contributed by Silicon Valley

will use the new funds to expand the size of the team, accelerate research and development, and establish new offices in Europe and the United States

According to the introduction, Mistral’s three co-founders will each sell more than 100 Ten thousand pounds of equity as part of the deal

This huge round of financing also confirms once again that investors are scrambling to find the "next OpenAI", and many large investors are willing to buy shares at high prices in the early stages of startups. , which is one of the reasons why startup valuations soar in their early stages.

The above is the detailed content of The business was valued at US$2 billion in 180 days! The European version of OpenAI explodes in popularity, Llama's creators set up their own business, and NVIDIA has become a shareholder. For more information, please follow other related articles on the PHP Chinese website!