Qwen 2.5 7B

Qwen 2.5 7B is an autoregressive text language model from Alibaba Cloud’s Qwen team. The cited checkpoint is a pretrained base model for causal text generation, not the instruction-tuned variant; the 7B scale makes it the compact Qwen 2.5 entry in this support matrix. It is not QwQ, the Qwen-family reasoning model, and it is not Qwen 3, the later generation. The name encodes both family version and approximate parameter size. TT-Metal lists Qwen 2.5 7B at TP=2 on an n300 Wormhole system with batch 32. In other words, one model is split across two accelerator devices, rather than using the eight-device QuietBox configuration listed for Qwen 2.5 72B.