UC Berkeley's NovaSky team has unveiled the Sky-T1-32B-Preview, an open-source AI model that matches OpenAI's o1-preview in math and coding capabilities. This model was developed in just 19 hours and at a cost of $450, utilizing eight Nvidia H100 GPUs. The rapid development and low cost highlight the potential for democratizing AI technology.
Sky-T1 was fine-tuned using Alibaba's Qwen2.5-32-Instruct and trained on data from QwQ-32B-Preview, showcasing innovative data preparation techniques. While it performs well on various benchmarks, it does not surpass OpenAI's o1 on more advanced physics-related questions. This development emphasizes the growing trend towards open-source AI solutions, which could level the playing field for smaller organizations.
• Sky-T1-32B-Preview matches OpenAI's o1-preview capabilities.
• The model was developed in 19 hours for $450.
Open-source AI refers to artificial intelligence models and tools that are freely available for use and modification, promoting accessibility and collaboration.
Synthetic training data is artificially generated data used to train AI models, helping to reduce costs and improve model performance.
Benchmarking in AI involves evaluating a model's performance against established standards or other models to assess its capabilities.
OpenAI is a leading AI research organization known for developing advanced AI models like o1-preview, which sets industry standards.
Nvidia is a technology company that produces GPUs, which are essential for training AI models efficiently, including the Sky-T1 model.
Business Insider on MSN.com 10month
Isomorphic Labs, the AI drug discovery platform that was spun out of Google's DeepMind in 2021, has raised external capital for the first time. The $600
How to level up your teaching with AI. Discover how to use clones and GPTs in your classroom—personalized AI teaching is the future.
Trump's Third Term? AI already knows how this can be done. A study shows how OpenAI, Grok, DeepSeek & Google outline ways to dismantle U.S. democracy.
Sam Altman today revealed that OpenAI will release an open weight artificial intelligence model in the coming months. "We are excited to release a powerful new open-weight language model with reasoning in the coming months," Altman wrote on X.