Insilico Medicine, a clinical-stage biotechnology company powered by generative artificial intelligence, has partnered with Liquid AI to introduce a new AI model designed to streamline pharmaceutical research and development. The collaboration aims to make advanced artificial intelligence more practical, secure, and accessible for scientists working to discover new medicines.

The partnership has produced LFM2-2.6B-MMAI, a compact yet powerful AI model capable of performing multiple drug discovery tasks within a single system. The model helps researchers predict molecular behavior, design new compounds, and optimize potential drugs while operating entirely on private company servers. This on-premise capability allows pharmaceutical organizations to protect sensitive data while benefiting from advanced AI-driven insights.

Unlike many AI systems used in drug development that rely on massive cloud-based models and large-scale computing infrastructure, LFM2-2.6B-MMAI takes a more efficient approach. With 2.6 billion parameters, the model can match or outperform AI systems up to ten times its size, significantly reducing computational requirements and enabling secure deployment within research laboratories.

According to Liquid AI CEO Ramin Hasani, the effectiveness of foundation models in scientific applications depends more on efficient architecture than on sheer size. He noted that the collaboration demonstrates how high-quality AI performance can be achieved while lowering the overall cost of computational intelligence.

The model assists scientists in answering critical questions throughout the drug discovery process, including identifying molecules that are likely to be safe and effective, improving compounds across multiple performance criteria, predicting molecular interactions with protein targets, and determining efficient laboratory synthesis routes. In testing, the system achieved up to 98.8 percent success in multi-parameter optimization, a key capability that helps chemists balance multiple drug design goals simultaneously.

The technology may be particularly valuable in research targeting aging-related diseases. Insilico Medicine focuses on therapeutic development for conditions such as fibrosis, neurodegenerative disorders, and other age-associated illnesses. By accelerating early-stage drug discovery, the company aims to shorten the timeline from scientific discovery to clinical treatment.

“Highly efficient liquid science models will make it easier for more scientists to achieve their goals, compress discovery timelines, and ultimately help patients,” said Alex Zhavoronkov, PhD, Founder and CEO of Insilico Medicine.

Beyond scientific impact, the partnership reflects a broader shift in pharmaceutical AI strategy. Instead of relying solely on increasingly large cloud-based models, the companies are prioritizing smaller, highly efficient systems that combine strong performance with improved data privacy and regulatory compliance.

If adopted widely across the pharmaceutical industry, the model could help accelerate the development of new therapies while strengthening Insilico Medicine’s position in AI-driven drug discovery. The ability to deploy powerful AI systems securely within company infrastructure may also appeal to large pharmaceutical organizations seeking both innovation and data protection.

Through this collaboration, Insilico Medicine and Liquid AI aim to transform how researchers approach drug discovery, making the process faster, more secure, and more accessible for scientists working to bring new treatments to patients worldwide.

New research from the UK Biobank suggests that elevated levels of the common amino acid tyrosine may be associated with reduced longevity, particularly in men.

Drawing on data from nearly 272,500 participants, researchers examined blood levels of the amino acids phenylalanine and tyrosine and tracked mortality outcomes over time. Approximately 24,000 deaths were recorded during the study period.

Using advanced genetic analysis, known as Mendelian randomization, the study assessed whether lifelong higher levels of tyrosine might play a causal role in shortening lifespan. The results indicate that genetically predicted higher tyrosine levels were associated with reduced longevity, with men losing roughly one year of life per standard deviation increase in tyrosine. The effect in women was weaker and less consistent.

“Tyrosine is a building block of protein and is essential for producing brain chemicals like dopamine, as well as stress-related hormones,” said the study authors. “Our findings suggest that the balance of specific amino acids, rather than total protein intake alone, may influence how we age.”

Phenylalanine, which can be converted into tyrosine in the body, did not independently predict lifespan once tyrosine levels were accounted for, positioning tyrosine as the primary factor in this association.

While the precise mechanisms remain under investigation, prior research suggests that amino acids like tyrosine may interact with nutrient-sensing pathways, cellular systems that regulate growth, repair, and stress responses, which could influence aging processes over decades.

The study also highlights a sex-specific difference, with men showing a stronger association between high tyrosine levels and shorter lifespan. Researchers caution that further studies are needed to understand these differences, including potential roles of hormonal regulation and metabolism.

Importantly, the study does not suggest eliminating protein from the diet or making abrupt dietary changes. Instead, it emphasizes the potential long-term impact of amino acid balance on health and aging.

This research marks a significant step toward understanding how microscopic biochemical factors, such as amino acids circulating in the blood over a lifetime, may shape human longevity and opens avenues for more personalized approaches to nutrition and aging interventions.