Qwen: Qwen3 Max Thinking

Qwen3-Max-Thinking is the flagship reasoning model in the Qwen3 series, specifically engineered for high-stakes cognitive tasks that demand deep, multi-step reasoning. By significantly scaling model capacity and leveraging advanced reinforcement learning compute, it achieves major gains in factual accuracy, complex reasoning abilities, and precise instruction following. This model also demonstrates strong alignment with human preferences and sophisticated agentic behavior, making it suitable for a wide range of demanding applications. This powerful text model offers a substantial context window of 262K tokens and can generate outputs up to 65K tokens. It supports advanced features such as JSON mode, streaming, function calling, and structured output. Access Qwen3-Max-Thinking via Multi AI's PRO tier for $1.20 per 1M input tokens and $6.00 per 1M output tokens, providing a cost-effective solution for advanced AI capabilities.