[Polish AI company Ingenix tops cellular senescence benchmark and open-sources model] According to TechFlow, on November 9th, Polish AI and biotechnology company Ingenix announced that its "Thalamus Reasoning" model ranked first among 29 models on LongevityBench, an independent benchmark for language models in cellular senescence research. The company also released an open-source version called "Thalamus Decision" for academic and commercial researchers to examine, reproduce, and further develop. According to the company, the model converts methylation, gene expression, and proteome data into mathematical vectors that the model can directly read (a process the company calls Modality Fusion). Each data file requires only 4 tokens, while a standard large model requires approximately 1.3 tokens to process a single word. The benchmark results are as follows: an average ranking of 1.42 across 14 ranking tasks (the second-best model scored 4.42, and the best-performing cutting-edge model, Claude Opus 4.6, scored 9.85), ranking first in 11 tasks; a DNA methylation age error of 3.36 years, compared to the best text-hinted model on the same Qwen3.5-9B pedestal of 12.9 years, also better than Horvath's clock at 4.67; and a transcriptome sorting accuracy of 0.828, compared to 0.671 for other best models. The company emphasizes that no benchmark data was used during training, eliminating data contamination. In terms of scale, this is a 9 billion parameter model, completed by a small team, and the company is developing a 95 billion parameter version. CEO Piotr Surma stated that leading cutting-edge labs and dedicated models on independent benchmarks validates the company's approach—allowing inference models to read biological data in its raw form, rather than reading text about it. Ingenix was founded in Warsaw in 2023. The team comes from the founding team of Applica, an AI company acquired by Snowflake in 2022. It had previously received €13 million in funding led by Sofinnova Partners.