👋 Welcome to ChemEAGLE Web
A multi-agent system for chemical information extraction. Upload a reaction or molecule image, or feed in a whole PDF paper — ChemEAGLE will detect the figures and return structured reactants, conditions, products and R-group expansions.
The built-in Ketcher editor on the right lets you instantly visualize, verify and fine-tune any extracted structure before exporting. Found something off? We'd love your feedback.
API Configuration
Select an API mode — click a card to switch.
⚡ Option B · GPT-5-mini Premium Active
Fill all three fields to use the GPT-5-mini version. Higher accuracy, requires an Azure OpenAI key.
Upload Image or PDF
Click or drag a file to upload
Supports PNG · JPG · PDF · or paste a screenshot (Ctrl+V) · or capture your screenExtracted Images from PDF
PDF Prediction Results
💡 Usage Guide · When to use which recognition mode? ▾
Start Recognition Image
Use this when you provide a single image (PNG / JPG). You can upload a file, drag & drop it onto the upload area, paste a screenshot from the clipboard (click 📋 Paste Screenshot or press Ctrl+V anywhere on the page), or directly capture your screen via 📷 Capture Screen — pick a screen / window / tab, then drag to crop the region you want before sending it for recognition. ChemEAGLE intelligently handles a wide range of chemistry figures — reaction schemes (with or without R-group / substituent tables), as well as pure molecule images (single or multi-molecule layouts). The full multi-agent pipeline runs end-to-end and returns structured results: reactants, conditions, products and R-group expansions for reactions, or recognized SMILES for standalone molecules. Best accuracy per figure — recommended whenever you already have a cropped image or a quick screenshot.
Extract Images and Text from PDF PDF
Use this when your input is a PDF paper and you want fine-grained control: ChemEAGLE first detects and crops every reaction figure from the PDF and shows them as thumbnails, and at the same time extracts the body text into a downloadable Markdown file. You can then review the figures and run recognition on individual ones, while keeping the full paper text for your own use. Recommended for long papers, supplementary information, or when only a few figures need processing.
Direct Predict PDF PDF
Use this for a one-click batch run on a PDF: ChemEAGLE will automatically detect every reaction figure and run full recognition on all of them, then return a consolidated result grid. Fastest end-to-end workflow, but consumes more compute / API quota — use it when you want everything in the document processed at once.
Recognition Results
Click any result below to send it to the editor →
