The weekend's AI news unfolded along two main lines: Qianwen launched a million-context full-modal model, and Grok upgraded its speech transcription capabilities; security and competitive boundaries also drew attention, with four major AI companies facing antitrust lawsuits for comments about slowing down collaborative research, and Google disclosing previous overreaching evaluations. On the tools side, OpenClaw updated its continuous operation mechanism, and Runway integrated web pages with the developer platform's points system. Model Pulse tracked changes in OpenRouter usage, Step 5 Preview third-party evaluations, and Grok transcription performance.