Qwen2.5-72B-Instruct is an instruction-tuned language model available on Hugging Face. It is designed to follow instructions, engage in dialogue, and handle tool use through structured function calling.
The model incorporates a specific prompt format that begins with a system message identifying it as Qwen created by Alibaba Cloud and positions it as a helpful assistant. When tools are supplied it includes an XML-based section that lists function signatures and instructs the model to return calls inside designated XML tags containing JSON objects with function name and arguments. The template supports both a chatml-style format with special tokens and a fallback variant depending on the presence of an initial system message.
It is delivered as a downloadable model repository on the Hugging Face platform. Developers can load and run the model using standard open-source inference libraries compatible with the repository format.
The model is provided under open-source terms through the Hugging Face hub.
Qwen2.5 72B Instruct sits in PulseGate's Foundation models & chat category. It focuses on enabling high-quality conversational AI and agentic capabilities with an open-weights model that can be self-hosted. It is built as an open-source project for developers. Qwen2.5 72B Instruct is open source under the Open Source license. Qwen2.5 72B Instruct is available on the web, the command line, and API.
It is developed by Qwen (China), and the product first shipped in 2024. Development happens publicly on GitHub with 27.4k stars. The category is crowded — PulseGate's index counts 25 comparable apps. Key capabilities include Instruction Following, Tool Calling, and Chat Format.
Latest indexed changes and source events
Qwen/Qwen2.5-72B-Instruct verified by the PulseGate indexer
Other apps tracked under the same category.
