Alibaba has released Qwen3.8-Max, a powerful 2.4 trillion parameter Mixture-of-Experts model. This model features a 1 million token context window and supports text, image, and video inputs. Its API is compatible with OpenAI and Anthropic protocols, easing integration for developers. Pricing is set at $2 per million input tokens and $6 per million output tokens. A significant cost-saving factor is the reduced price of cached input tokens, emphasizing the importance of stable prefixes in prompts.The model's naming convention distinguishes generations from point releases, with Qwen3.8-Max being the latest flagship. While it boasts 2.4 trillion total parameters, only approximately 95 billion are active per token, making inference more efficient. The effective context window is around 991K tokens, with a maximum output token limit of 131K. Developers can utilize its OpenAI-compatible API by updating the base URL and model name.Alibaba's DashScope SDK provides an example of its multimodal capabilities with a code snippet. The model supports various features including function calling and structured outputs, and comes with five built-in tools. Benchmarks show strong performance in multimodal and agentic tasks, though some areas lag behind competitors like Claude 3.5. Open weights are expected to be released soon, along with a smaller 27B parameter version.Currently, a formal model card with detailed training data and safety evaluations is missing. The licensing terms for commercial use will only be clear once the open weights are released. The active parameter count is reported but not yet officially confirmed by Alibaba. Despite these missing pieces, Qwen3.8-Max is recommended for multimodal applications, long-context tasks, and existing OpenAI/Anthropic protocol users.