SpecXMaster enables chemists to determine molecular structures from experimental NMR data, returning ranked candidate structures together with a detailed elucidation report. In a head-to-head evaluation of 500 structure-elucidation cases, SpecXMaster correctly solved 77.8%, substantially outperforming leading general-purpose AI models (Gemini 3.1 Pro Preview at 44.8% and Claude Opus 4.7 at 35.0%).
Most analyses can be completed within three minutes, making the platform useful for identifying unknown compounds, distinguishing closely related candidates, and revisiting questionable structural assignments. In our real research case, SpecXMaster identified within about one minute a side-product structure that had previously required weeks of additional experiments and spectral reanalysis. As of July 2026, the platform had served more than 2,000 users across over 10,000 conversations (collaboration with DP, AISI).
