Mistral AI CEO Arthur Mensch announced on French National Radio that Mistral AI will release an open-source GPT-4 level model in 2024. Mistral recently unveiled the 8X7B model based on the MoE architecture, which is comparable to other popular models such as GPT 3.5 and Llama 2 70B. Mistral outperforms Llama 2 70B in most benchmark tests and has a 6x faster inference speed. It is licensed under Apache 2.0. The Paris-based startup recently announced that it has raised $415 million, valuing the company at $2 billion. Andreessen Horowitz (a16z) led the latest funding round, with Lightspeed Venture Partners also participating. Maintaining business operations is often challenging for open-source large language model (LLM) companies. To overcome this, Mistral AI has launched a platform called "La Plateforme," where it will provide API endpoints for its existing models. The company has created three categories for its models - Mistral Tiny, Mistral Small, and Mistral Medium. Mistral 7B Instruct v0.2 and Mixtral 8x7B belong to Mistral Tiny and Mistral Small, respectively. Interestingly, the Medium model has not been released yet. Mistral AI is currently developing the Mistral Medium model, which ranks among the top service models in standard benchmark tests. It excels in English, French, Italian, German, Spanish, and coding, with a score of 8.6 on MT-Bench. Theoretically, it even surpasses GPT 3.5. Meanwhile, there are rumors that OpenAI may release the GPT-4.5 version before the end of December.