Meta working on new AI model more powerful than ChatGPT 4, does not want to rely on Microsoft for training

Meta is reportedly developing a new AI model that could surpass OpenAI's most advanced model, GPT-4. The company is also planning to train its AI system independently, reducing its reliance on Microsoft. Meta's AI system, set to begin training in early 2024, is expected to be more powerful than GPT-4, which currently powers ChatGPT's premium version and is capable of writing code, composing poetry, and solving complex problems.

ChatGPT 4, introduced in March 2023, is a large multimodal model that accepts image and text inputs and emits text outputs. It has demonstrated human-level performance on various professional and academic benchmarks, including the LSAT, SAT Math, and GRE exams. The exact capabilities of Meta's new AI chatbot compared to ChatGPT 4 remain to be seen.

Key takeaways

Meta is reportedly developing a new AI model that could potentially be more powerful than OpenAI's most advanced model, GPT-4.
The company is also taking steps to ensure it doesn't have to rely on Microsoft for training its AI system, including purchasing more of Nvidia's H100 AI-training chips and improving its infrastructure.
Meta's new AI model is expected to begin training in early 2024 and aims to outperform OpenAI's GPT-4, which is currently the most sophisticated generative AI chatbot.
ChatGPT 4, introduced in March 2023, is a large multimodal model that has demonstrated human-level performance on various professional and academic benchmarks, including LSAT, SAT Math, and GRE exams.

Meta working on new AI model more powerful than ChatGPT 4, does not want to rely on Microsoft for training

Key takeaways

Discussion (0)