{"slug": "retijev-fast-and-source-grounded-ophthalmic-decision-models", "title": "RetiJev: Fast and Source-Grounded Ophthalmic Decision Models", "summary": "Researchers released RetiJev, a family of 0.8B-9B decision models that score candidate answers directly without autoregressive decoding, built on RetiJev-Data, a source-grounded ophthalmology corpus of 78,158 clinical-decision and diagnosis questions drawn from PubMed Central articles and a MeSH vocabulary covering 531 eye diseases, plus 50,074 guideline-grading questions spanning 11 grading systems and a 24,662-question evaluation benchmark. Across all model sizes, RetiJev outperformed both a 27B LLM and the commercial Jev 1.13 decision model on guideline grading by 10.4-23.0 points, and RetiJev-9B achieved the highest accuracy on clinical decisions, improving over both baselines by 5.4-6.5 points while matching them on diagnosis. RetiJev answers a case in 158 ms on one GPU, 3.5x faster than LLM generation and 470x faster than reasoning, with the models released at github.com/Awenbocc/RetiJev.", "body_md": "arXiv:2610.30947v1 Announce Type: new \nAbstract: Ophthalmic clinical decision support requires fast, accurate, and privacy-preserving inference, yet autoregressive large language models (LLMs) incur substantial latency and computational overhead for decisions among predefined actions. We therefore investigate a decision-oriented alternative to generative reasoning. Specifically, we first introduce RetiJev-Data, a source-grounded ophthalmology corpus comprising 78,158 clinical-decision and diagnosis questions derived from articles in PubMed Central and a MeSH vocabulary covering 531 eye diseases, and 50,074 guideline-grading questions spanning 11 grading systems, along with a 24,662-question evaluation benchmark. Based on this, we develop RetiJev, a family of 0.8B-9B decision models that directly score candidate answers without autoregressive decoding. Across all model sizes, RetiJev outperforms both a 27B LLM and the commercial Jev 1.13 decision model on guideline grading by 10.4-23.0 points. RetiJev-9B further achieves the highest accuracy on clinical decisions, improving over both baselines by 5.4-6.5 points while matching them on diagnosis. It answers a case in 158 ms on one GPU, 3.5x faster than LLM generation and 470x faster than reasoning. Models are released at https://github.com/Awenbocc/RetiJev.", "url": "https://wpnews.pro/news/retijev-fast-and-source-grounded-ophthalmic-decision-models", "canonical_source": "https://www.machinebrief.com/news/retijev-fast-and-source-grounded-ophthalmic-decision-models-zl50", "published_at": "2026-10-23 04:00:00+00:00", "updated_at": "2026-10-23 04:42:54.817446+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-research"], "entities": ["RetiJev", "RetiJev-Data", "RetiJev-9B", "Jev 1.13", "PubMed Central", "MeSH", "GitHub"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/retijev-fast-and-source-grounded-ophthalmic-decision-models", "markdown": "https://wpnews.pro/news/retijev-fast-and-source-grounded-ophthalmic-decision-models.md", "text": "https://wpnews.pro/news/retijev-fast-and-source-grounded-ophthalmic-decision-models.txt", "jsonld": "https://wpnews.pro/news/retijev-fast-and-source-grounded-ophthalmic-decision-models.jsonld"}}