On October 21, 2025, Anthropic announced a series of major upgrades to its AI model Claude, with the core focus on enhancing its application capabilities in the life sciences field. The goal is to accelerate the R&D speed in this sector by an order of magnitude and achieve the ambitious target of “accomplishing 100 years of scientific progress in 10 years”. Meanwhile, Anthropic launched the web version of its AI programming tool Claude Code and revealed a multi-dimensional layout for Claude to evolve into a “superhuman research assistant”.
The core support for this upgrade comes from the latest generation model, Claude Sonnet 4.5, which is Anthropic’s first model trained systematically in scientific knowledge and has delivered impressive performance in multiple life science benchmark tests. In the Protocol QA test, which assesses the ability to understand and apply laboratory protocols, it scored 0.83, surpassing not only the human baseline score of 0.79 but also significantly outperforming the previous generation model’s 0.74. It also achieved substantial improvements in BixBench, an evaluation for bioinformatics tasks, laying a solid technical foundation for its deep engagement in the life sciences field.
To enhance scientific practicality, Anthropic has advanced functional upgrades in three key directions. Firstly, it launched a series of scientific platform connectors covering mainstream professional tools and databases such as Benchling, BioRender, and PubMed. These connectors enable Claude to link experimental records for traceability, generate compliant scientific charts, access a vast amount of biomedical literature, support data sharing and analysis as well as complex bioinformatics analysis. They complement existing general-purpose tools and data analysis platforms, integrating deeply into the research workflow.
Secondly, it introduced the “Agent Skills” feature. Existing as preset packages containing instructions, scripts, and resources, this feature ensures the consistency and predictability of Claude’s operations when performing specific tasks. Anthropic is developing the first batch of scientific skills, such as the automated quality control and filtering skill for single-cell RNA sequencing data based on scverse best practices, while encouraging scientists to customize skills to further expand application scenarios.
Thirdly, it built a dedicated prompt library for life sciences. The upgraded Claude is capable of undertaking diverse tasks including literature reviews, hypothesis formulation, research proposal drafting, genomic data analysis, and regulatory document review. The dedicated prompt library will help users quickly master operation methods and improve efficiency.
Notably, the concurrently launched web version of Claude Code breaks free from the limitations of local terminal environments. Users can entrust programming tasks directly through a browser, and it also supports cloud-based parallel development, real-time progress tracking, and secure sandbox operation, significantly lowering the barrier to use. Currently, Claude Code has transcended pure programming applications and is widely used in the life sciences field for tasks such as paper drafting, literature reviews, and research project management, as part of the latest AI news in the tech and scientific communities.
Anthropic stated that it aims for Claude to cover the entire lifecycle of life sciences, from early-stage research and translation to commercialization, serving as a capable assistant for research professionals. Many customers and partners are already applying it to practical scientific tasks, and the company is also providing free API credits to leading laboratories through its “AI for Science” program to support cutting-edge research and explore new application scenarios. This upgrade marks that following its foray into the programming field, Claude is officially delving into the vertical track of life sciences, demonstrating the enormous potential of deep integration between AI and vertical sectors.
