Ingenix Publishes Benchmarks for Thalamus Decision, World’s-First Typed-Decision Biological Model Capable of Interpreting Gene-Expression Data
WARSAW, Poland, Nov. 17, 2026
Press Release Disclaimer: This is a press release distributed through the XPR Media network. It has not been independently verified by our newsroom.

![]()
Ingenix Publishes Benchmarks for Thalamus Decision, World’s-First Typed-Decision Biological Model Capable of Interpreting Gene-Expression Data
PR Newswire
WARSAW, Poland, Nov. 17, 2026
- Thalamus Decision makes calibrated, single-pass decisions on biological samples.
- At 9B and 27B parameters, it outperforms text-only medical decision models by 37 points or more on sample-grounded biology tasks.
- Thalamus Decision is capable of automating almost two-thirds of early-stage drug development decisions.
WARSAW, Poland, Nov. 17, 2026 /PRNewswire/ — Polish AI start-up Ingenix has today published benchmarks for Thalamus Decision, which to the company’s knowledge is the world’s first typed decision model for biology. Instead of writing out an answer like an LLM, typed decision models produce confidence scores for every possible option, so systems know when to act on a decision and when to ask a human.
Until now, typed decision models could only read text and, in some cases, images. Thalamus Decision can read bulk and single-cell RNA-seq expression profiles, and is available at two sizes, 9B and 27B parameters.
Built on Ingenix’s Thalamus Reasoning (which recently claimed first place out of 29 other models on LongevityBench), Thalamus Decision handles tasks including assigning cell type, judging cell malignancy, identifying a sample’s tissue of origin and predicting cell lines’ drug response. It can also assess perturbation response – whether silencing a gene changes expression, and in which direction, a critical early-stage process required for genetic medicine IND studies.
For drug discovery, Thalamus Decision scored 84–89% on two GDSC cell-line drug-response tasks, approximately 94% on gene-essentiality across tissues, and approximately 95% when identifying a cancer model’s type (cell line, organoid, PDX or patient tumour). Calibrated confidence lets the 9B model automate 62.6% of decisions at a 5% target error rate, with a 6.3% realised error rate. For the 27B model 66.2% of decisions were automated, with a 6.8% realised error rate.
For drug developers, this suggests that Thalamus Decision could be used to automate almost two-thirds of early-stage drug development decisions, with only around 6% of these being incorrect. This would vastly accelerate the pace of drug development.
Speaking on the findings, Ingenix’s CEO, Piotr Surma said: “Drug discovery is generating more promising molecules and proteins than ever before, across oncology, CNS, inflammatory disease and many other areas. The bottleneck is increasingly deciding which candidates to prioritise and take into the lab. Thalamus Decision is designed to help make those decisions, using biological data to identify the candidates most worth progressing towards experimental validation and ultimately IND-enabling studies. We’re incredibly excited about the role it could play.”
The 9B version of Thalamus Decision will be open-sourced on November 18.
About Ingenix
Ingenix was founded around a single question: what would it take for AI to truly understand biology? The Ingenix Biological Reasoning Engine is the company’s answer. Built on Modality Fusion, a novel AI architecture that integrates best-in-class models across modalities and biological scales, then reasons across their representations directly. Ingenix is the first to name this approach and the first to bring it to biology. The engine is built for translational and clinical R&D decisions, and every output arrives with the reasoning trace behind it: mechanism, evidence, and the caveats laid out alongside the conclusions. Founded in 2023 and headquartered in Warsaw, Ingenix is led by AI researchers and engineers from the founding team of Applica, the AI company acquired by Snowflake in 2022, whose TILT model contributed to the GPT-4 benchmark. For more information, visit ingenix.ai.
Media Contact
Barnaby Pickering, Director, 59 North Communications
Barnaby.Pickering@59north.bio
View original content to download multimedia:https://www.prnewswire.com/news-releases/ingenix-publishes-benchmarks-for-thalamus-decision-worlds-first-typed-decision-biological-model-capable-of-interpreting-gene-expression-data-302941005.html
SOURCE Ingenix

