OpenAI has introduced GPT-Rosalind, a frontier reasoning model designed to accelerate research across biology, drug discovery, and translational medicine. This model is optimized for scientific workflows, combining improved tool use with a deeper understanding of chemistry, protein engineering, and genomics.

Drug development typically takes 10 to 15 years, with complex research workflows often hindering progress. GPT-Rosalind aims to help researchers navigate the early stages of discovery faster by supporting evidence synthesis, hypothesis generation, and experimental planning in multi-step research tasks.

The model is available as a research preview in ChatGPT, Codex, and the API for qualified customers through a trusted access program. OpenAI has also introduced a freely accessible Life Sciences research plugin for Codex, enabling scientists to connect models to over 50 scientific tools and data sources.

OpenAI is collaborating with customers such as Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific to apply GPT-Rosalind across workflows that accelerate research and discovery. The model is named after Rosalind Franklin, whose research was crucial in revealing the structure of DNA.

Evaluations show that GPT-Rosalind delivers superior performance in scientific subdomains including chemistry, biochemistry, protein understanding, phylogenetics, experimental design, and tool usage. It has achieved leading results on benchmarks like BixBench and LABBench2, outperforming other models.

Access to the model is provided to scientists and research organizations best positioned to advance human health, while maintaining strong safeguards against biological misuse. Access is evaluated based on three core principles: beneficial use, strong governance and safety oversight, and controlled access with enterprise-grade security.

👉 Read the full article on OpenAI

Physical Intelligence, a rapidly growing robotics startup, has unveiled a new robot brain capable of autonomously performing tasks it was never explicitly taught. This technology is expected to significantly enhance robots' ability to understand their environment and solve novel problems, even in unprogrammed situations.

The company's robot brain leverages principles of Generative AI and AI Agents to enable reasoning and action planning in complex physical environments. This could accelerate the adoption of robots across various industries, including manufacturing, logistics, and service sectors.

This breakthrough signifies that robots can become more versatile and adapt to unforeseen circumstances without human intervention. It is considered a significant step forward in the field of robotics, potentially moving closer to Artificial General Intelligence (AGI).

Physical Intelligence emphasizes that this new robot brain overcomes the limitations of flexibility faced by traditional robots, allowing them to operate in more complex and dynamic settings.

👉 Read the full article on TechCrunch

OpenAI has announced a significant update to Codex, enhancing its control capabilities over desktop environments, a move aimed at competing with rival Anthropic. This beefed-up Codex will be able to automate and execute a wider range of tasks directly on a user's desktop.

The new Codex expands its functionality as an AI Agent, capable of performing complex, multi-step operations based on natural language instructions, such as inter-application coordination, file manipulation, and system setting changes. This is expected to dramatically boost productivity for both developers and general users.

This strategic move indicates OpenAI's direct competition with advanced agent capabilities offered by models like Anthropic's Claude, aiming to establish leadership in the desktop AI assistant market.

The enhanced Codex will provide more powerful LLM-based automation not only for programming tasks but also for everyday desktop operations, helping users efficiently manage more complex workflows.

👉 Read the full article on TechCrunch