{"slug": "deeplens-diagnosis-agent-agentic-workflow-design-lets-a-small-reasoning-model", "title": "DeepLens Diagnosis Agent: Agentic Workflow Design Lets a Small Reasoning Model Compete with Frontier LLMs", "summary": "A new AI agent called DeepLens Diagnosis Agent, designed by researchers and centered on the small medical reasoning model JSL Medical Small 7B v2, achieved 60.14% top-1 diagnostic accuracy on the 915-case DiagnosisArena benchmark, outperforming frontier LLMs like Claude Sonnet 4.5 and Gemini 3.1 Pro while costing USD 0.0072 per case. The agent's five-stage pipeline, which enforces structured clinical extraction, retrieval-augmented generation, and evidence triangulation, improved accuracy by 36 percentage points over the same model without the workflow, demonstrating that process constraints can compensate for model size and cost.", "body_md": "arXiv:2607.22555v1 Announce Type: new\nAbstract: Medical diagnosis is a multi-stage process: extract facts, consult knowledge, generate a differential analysis, and select the best diagnosis with explanations. Frontier LLMs are strong generalists, but single-shot prompting often yields brittle diagnostic reasoning. We present the DeepLens Diagnosis Agent, a five-stage harnessing pipeline (combining model capabilities with disciplined process constraints) centered on a small medical reasoning model (JSL Medical Small 7B v2) and retrieval-augmented generation (RAG). The pipeline enforces structured clinical extraction, disciplined retrieval, constrained candidate generation, explicit evidence triangulation, and an auditable final decision. On the 915-case DiagnosisArena benchmark, the agent achieved 60.14% top-1 diagnostic accuracy, the highest among small and medium-sized models. The same model without the agent workflow achieved 23.99%, a +36-point gain from workflow design alone, despite 88.2% on standard medical benchmarks, showing that diagnostic reasoning under uncertainty requires more than knowledge recall. The agent costs USD 0.0072 per case (24K tokens on A100) with 24-second latency, 35-45% cheaper than Claude Sonnet 4.5 (USD 0.0110) and Gemini 3.1 Pro (USD 0.0128) while outperforming them by +9.70pp and +9.17pp. Harnessing can also correct frontier model failures; workflow constraints can outweigh parameter count or API cost.\nBeyond aggregate accuracy, the pipeline produces structured intermediate artifacts that make each stage inspectable and support error localization. These properties support high-stakes settings where traceability, reproducibility, and auditable evidence matter alongside benchmark performance.", "url": "https://wpnews.pro/news/deeplens-diagnosis-agent-agentic-workflow-design-lets-a-small-reasoning-model", "canonical_source": "https://arxiv.org/abs/2607.22555", "published_at": "2026-07-28 04:00:00+00:00", "updated_at": "2026-07-28 04:28:44.637488+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-research", "ai-products"], "entities": ["DeepLens Diagnosis Agent", "JSL Medical Small 7B v2", "DiagnosisArena", "Claude Sonnet 4.5", "Gemini 3.1 Pro"], "alternates": {"html": "https://wpnews.pro/news/deeplens-diagnosis-agent-agentic-workflow-design-lets-a-small-reasoning-model", "markdown": "https://wpnews.pro/news/deeplens-diagnosis-agent-agentic-workflow-design-lets-a-small-reasoning-model.md", "text": "https://wpnews.pro/news/deeplens-diagnosis-agent-agentic-workflow-design-lets-a-small-reasoning-model.txt", "jsonld": "https://wpnews.pro/news/deeplens-diagnosis-agent-agentic-workflow-design-lets-a-small-reasoning-model.jsonld"}}