Model Positioning
Qwen2.5-Max is a large-scale mixture-of-experts (MoE) language model developed by Alibaba's Qwen team. According to the official announcement, it was pretrained on over 20 trillion tokens and post-trained using supervised fine-tuning and reinforcement learning from human feedback. (qwenlm.github.io)
Capabilities and Uses
The model is designed for text understanding and generation and can be used for tasks such as multilingual dialogue, text writing, code generation, and logical reasoning. The official model information page states that the rolling version qwen-max is functionally equivalent to the snapshot qwen-max-2025-01-25; specific feature support varies by region. (alibabacloud.com)
Limitations and Availability
Model outputs may contain errors; users should verify important information and code. The official qwen-max information page lists a context window of 32,768 tokens, a maximum input of 30,720 tokens, and a maximum output of 8,192 tokens. These are the limits listed on the page for the rolling version and should not be regarded as guarantees for all deployments. (alibabacloud.com)
In April 2026, Alibaba Cloud announced plans to decommission historical models, including this snapshot, in the Beijing and Singapore regions starting at 00:00 on May 13, 2026 (UTC+8), subject to the actual decommissioning time. Actual availability in the target region should be confirmed before use. (alibabacloud.com)
Release Date
January 28, 2025 is the date of the official Qwen announcement, which introduced Qwen2.5-Max and announced API availability for model qwen-max-2025-01-25. The date recorded here is based on the official announcement and availability event; 2025-01-25 in the model ID is not treated as the initial release date. (qwenlm.github.io)
Source note: In the current data, whether this snapshot was first released or made available before January 28, 2025 has not been independently verified. The date above is based on the date and API availability statement in the official announcement. (qwenlm.github.io)
Pricing
The following prices are listed on official pages, all queried on November 20, 2026; prices for different platforms and models cannot be treated as belonging to the same plan.
- CNY, Alibaba Cloud Enterprise Agent Application Platform AIpaas: For
qwen-max-2025-01-25, input is CNY 0.0024 per thousand tokens and output is CNY 0.0096 per thousand tokens. The page bills input and output separately according to actual token usage; actual API availability must be confirmed separately. Official billing rules. (help.aliyun.com) - USD, Alibaba Cloud International Model Studio, Singapore region, international deployment: For the rolling model
qwen-max, the original price is USD 1.6 per million input tokens and USD 6.4 per million output tokens. According to the official page, the rolling version is functionally equivalent to this snapshot, but the price is listed forqwen-maxand is not a guarantee of current pricing for the decommissioned snapshot ID. Official model and pricing page. (alibabacloud.com)