Qwen4-Max ranks top-5 on DesignArena PPTX and spreadsheet leaderboards
The 379-point Elo leap on spreadsheets suggests a substantial architectural or training improvement, not an incremental update, in a benchmark area few vendors optimise for explicitly. Enterprise developers building office-automation and document-pipeline tools should compare Qwen4-Max against Anthropic and OpenAI offerings on these specific workloads before defaulting to established choices. Against Kimi's top models, Chinese labs now compete across both chat and deliverable-generation benchmarks.