New benchmark study results show leading AI models, including ChatGPT, Claude, and Gemini, still lag humans in visual math reasoning.
After years of creating highly specialized software, researchers used supercomputer clusters to finally solve the "100,000-body problem.