# CERTA Tests When Smaller Clinical AI Models Can Be Used

> Researchers evaluated a method that lets quantized models handle clinical-verification decisions when they match a BF16 reference closely enough, escalating the rest.

Oossa · 2026-10-22 · https://oossa.com/en/certa-tests-when-smaller-clinical-ai-models-can-be-used

Researchers introduced CERTA, a method for deciding when a compressed clinical language model can handle a verification decision instead of a larger BF16 model. Decisions that fail CERTA’s statistical check are sent to the reference model.

In tests across 11 models and 2,500 decisions, 8-bit versions handled 99.8% to 100% of held-out decisions in the study’s operational analysis. For two 72-billion-parameter models, selected 4-bit configurations had estimated deployment costs around 26% to 27% of the BF16 reference and peak memory use below 48 GB. The researchers did not test deployment on that specific hardware, and the results concern a clinical-verification task, not clinical care generally.

## The facts

- The evaluation used 2,500 decisions from contexts involving 2,484 patients, with patient-disjoint calibration and testing.
- For Qwen2.5-72B and HuatuoGPT-o1-72B, selected 4-bit configurations used 43.0 GB and 42.7 GB peak memory, respectively; deployment on a 48 GB accelerator was not directly tested.

## Why it matters

For teams exploring local clinical-verification systems, CERTA suggests a way to route some decisions to cheaper compressed models, but the study does not establish safety or performance for other clinical uses or real-world deployment.

## Sources & references

1. [Do We Need the Largest LLMs for Every Clinical Decision?](https://arxiv.org/abs/2610.26937) – arXiv, 2026-10-21

Last updated: 2026-10-22