Qwen 2.5 Released with Enhanced Reasoning Capabilities
Summary
Alibaba Cloud announces Qwen 2.5, featuring significant improvements in mathematical reasoning, code generation, and multilingual support. The model achieves state-of-the-art performance on multiple benchmarks.
Alibaba Cloud today announced the release of Qwen 2.5, the latest iteration of its large language model series. This release brings substantial improvements across multiple dimensions, with particular strength in mathematical reasoning, code generation, and multilingual understanding.
Qwen 2.5 achieves new state-of-the-art results on MMLU, HumanEval, and GSM8K benchmarks, positioning it among the top open-source models available. The model is available in multiple sizes (7B, 14B, 32B, 72B) and supports context windows up to 128K tokens.
Developers can access Qwen 2.5 through Alibaba Cloud Model Studio or download weights from Hugging Face.