Poe API

Qwen3-Max-Thinking

Qwen3-Max-Thinking is a flagship reasoning model that integrates adaptive tool use, autonomously employing search, memory, and code interpretation to address complex tasks. It further optimizes performance through test-time scaling, a strategy that allocates additional computation during inference to improve reasoning accuracy and context efficiency. This model is served by Alibaba Cloud Int. from Singapore. Notes: - Context window: 256K - Supported file types: PNG, JPG, JPEG, HEIC, PDF, TXT, DOC, DOCX - Audio/Video files are not supported. Parameter controls available: 1. Model Configuration Thinking to let the model reason step-by-step before answering. It is `on` by default. 2. Tools - Web search to search the web for real-time info. It is `on` by default. - Web search mode. Select from `Standard` for efficient search or `Thorough` for comprehensive search. It is set to `Standard` by default (requires thinking enabled). - Web extractor to extract/read content from URLs (requires thinking enabled). It is `on` by default. - Code interpreter to run Python code in a sandbox (requires thinking enabled). It is `on` by default.