Biology AI Model Tops Aging Benchmark in Company Test
Ingenix's 9-billion-parameter model reports a first-place LongevityBench ranking, a result not yet independently verified.
Updated on Nov. 9, 2026 in Life Sciences

Warsaw-based Ingenix says its 9-billion-parameter Thalamus Reasoning model ranks first out of 29 models on LongevityBench, a benchmark of 14 ranked tasks. The result is a company-reported benchmark outcome, not an independently verified or peer-reviewed one, and the model is not described as a shipped consumer product.
Why it matters
The claim sits at the intersection of large language models and biological data, where general-purpose text models have so far struggled with molecular measurements. If the ranking holds up under independent evaluation, it suggests that models built around multi-omic inputs can outperform text-prompted systems on aging-related tasks.
Thalamus Reasoning posted a transcriptomics ordering accuracy of 0.828 and a DNA methylation age error of 3.36 years, against 12.9 years for the best text-prompted model on the same base. The company states that no benchmark data was used in training, though the extraction does not say whether these results have been independently reproduced.
The players
Ingenix
Warsaw-based company building the Biological Reasoning Engine, which includes the 9-billion-parameter Thalamus Reasoning model and a 95-billion-parameter model in development.
The details
Thalamus Reasoning converts methylation, gene expression and proteomics measurements into mathematical vectors rather than treating them as text. Ingenix calls its process for generating each biological profile Modality Fusion, which requires four tokens. Ingenix describes the model as part of its Biological Reasoning Engine.
Timeline
November 9, 2026: Ingenix announces the LongevityBench ranking and open-sources Thalamus Decision, a version of the model.
The Tech Race
The result places a biology-specific reasoning model against general-purpose frontier systems such as Claude Opus 4.6, which posts a mean rank of 9.85 on the same benchmark. It is one entrant in a growing effort to build foundation models that read molecular data directly rather than through text descriptions.
Thalamus Decision, a version of the model, is slated for open-source release on November 9, 2026, letting researchers test the approach directly. No consumer availability, pricing or clinical deployment timeline has been announced.
The takeaway
Watch whether the November 9, 2026 open-source release of Thalamus Decision allows independent researchers to reproduce the LongevityBench ranking. Independent verification, rather than the company's reported 1.42 mean rank, will determine whether the result holds.
Further reading
For broader context on how machine learning is being applied to biological data, explore the Life Sciences section.


