MedJev: A System One Vision-Language Decision Model for Fast and Effective Medical Image Understanding MedJev introduces a System One vision-language decision model for medical image understanding that avoids autoregressive reasoning and answer-token generation, which the authors say makes inference cost scale with response length and forces answers to be parsed from free-form text. The model is presented as a faster, more effective alternative to conventional VLM pipelines for medical imaging. The work targets latency and parsing overhead in clinical image analysis. Medical image understanding with vision-language models VLMs typically relies on autoregressively generating reasoning and answer tokens. These unnecessary tokens make the inference cost scale with the response length and force the answer to be parsed from free-form text, incurring avoidable laten