Researchers at Heidelberg University and the German Cancer Research Center in Germany have developed Hetairos, an artificial intelligence system capable of classifying central nervous system tumors directly from standard histological slides. Published in Nature Cancer, the technology reduces diagnostic turnaround from approximately twelve days to twelve minutes while maintaining accuracy comparable to expert neuropathologists. Traditional brain tumor diagnosis relies on DNA methylation analysis, which serves as the clinical gold standard but requires specialized laboratories, costly equipment, and ample tissue samples. The process typically delays treatment planning by two weeks and remains inaccessible in many resource-limited regions. Hetairos addresses these bottlenecks by predicting molecular subgroups exclusively from routinely stained microscopic sections. The model was trained and validated on a dataset exceeding eleven thousand digitized tissue slides from 9,606 patients across eleven medical centers on four continents. It identifies 102 distinct molecular tumor subtypes aligned with the current World Health Organization classification. Performance testing demonstrates the system clinical viability. In head-to-head evaluations involving 210 cases, Hetairos achieved a primary diagnosis accuracy of 68 percent, significantly outperforming five seasoned neuropathologists who averaged 30 percent accuracy. When evaluating the top three likely classifications, the AI reached 84 percent compared to approximately 50 percent for human experts. The system also outputs confidence metrics, delivering high-certainty results in 50 to 70 percent of cases with 87 to 88 percent precision. In lower-confidence scenarios, Hetairos effectively narrows differential diagnoses to a manageable subset, streamlining subsequent laboratory testing. Prospective clinical trials confirm operational efficiency. After initial tissue digitization, the algorithm processes samples in roughly twelve minutes on standard computing hardware. End-to-end results typically become available within one to two days. The model also highlights relevant tissue regions, providing neuropathologists with transparent decision support and guidance for targeted molecular verification. While the AI faces challenges with exceptionally rare tumor variants, researchers project performance improvements as training datasets expand. Lead developers Moritz Gerstung and Felix Sahm emphasize that Hetairos is designed as a diagnostic adjunct rather than a replacement for molecular profiling. The approach eliminates per-test fees associated with methylation arrays, offering substantial cost reductions and democratizing precision oncology in regions lacking advanced genetic infrastructure. By leveraging globally available histopathology workflows, the technology establishes a new benchmark for rapid, accessible cancer classification. The research underscores the growing capacity of digital pathology AI to extract complex molecular insights from conventional morphological data, fundamentally accelerating neuro-oncology diagnostics worldwide.