The Builder's Brief
OpenAI's $122B funding, APEX-Agents dev set results
Friday, May 29, 2026
OpenAI has secured $122B in funding to develop its superapp, which will integrate various AI models and tools. This investment is expected to accelerate AI research and development, with potential applications in industries such as healthcare and finance. The funding will also support the development of more advanced AI models, including those capable of complex reasoning and decision-making.
Read the full story on therundown.aiAC-Small has shown significant improvement on held-out benchmarks after post-training on the APEX-Agents dev set, with +5.7pp on APEX, +8.0pp on Toolathalon, and +7.7pp on GDPval. This demonstrates the potential of fine-tuning AI models on specific datasets to achieve state-of-the-art performance. The results also highlight the importance of evaluating AI models on a range of benchmarks to ensure their robustness and generalizability.
Read the full story on mercor.comOpenAI has reportedly secured a $1B investment from Disney, which will support the development of AI-powered entertainment and media content. This partnership is expected to drive innovation in the entertainment industry, with potential applications in areas such as virtual reality and interactive storytelling. The investment also highlights the growing interest in AI from major media companies.
Read the full story on therundown.aiA recent article suggests that AI is not replacing jobs, but rather 'unbundling' them into smaller, lower-paid tasks. This phenomenon has significant implications for the future of work and the economy, as it may lead to increased inequality and decreased job security. The article highlights the need for policymakers and business leaders to consider the impact of AI on the workforce and to develop strategies for mitigating its negative effects.
Read the full story on theregister.comResearchers have proposed SCoOP, a training-free uncertainty quantification framework for multi-VLM systems. SCoOP uses uncertainty-weighted linear opinion pooling to combine the outputs of multiple Vision-Language Models, which can enhance multimodal reasoning and robustness. The framework has potential applications in areas such as computer vision and natural language processing.
Read the full story on arXivA new study explores the concept of 'graceful forgetting' in generative language models, which refers to the ability of models to selectively forget certain knowledge or information. The researchers propose a method for achieving graceful forgetting, which can improve the performance and efficiency of language models. The study has implications for the development of more advanced language models and their applications in areas such as language translation and text summarization.
Read the full story on arXivA recent article argues that agentic AI systems need boundaries and constraints before they can be granted freedom. The author suggests that without such boundaries, AI systems may pose significant risks to humans and the environment. The article highlights the need for careful consideration of the ethical implications of AI development and the importance of establishing clear guidelines and regulations for the development and deployment of AI systems.
Read the full story on news.google.com• Google releases AI-powered coding tool
• Microsoft invests in AI startup
• AI-generated art sells for $1M
Think your CV is perfect? Let AI prove you wrong in seconds — brutal, honest, and hilarious feedback.
Try it free →🧠 Fun Fact: 45% of AI models use transfer learning
Get this in your inbox every week
Join builders who start their day with The Builder's Brief
Subscribe Free