Qwen2Open Weight
Released
Qwen2 is an AI model released by Qwen on Thursday, Jun 6 2024, 123 days after Qwen1.5. It is an open-weight model — the trained weights are available to download and run. It was released in 0.5B and 1.5B and 7B and 57B and 72B parameter sizesThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality..
Get Qwen releases and AI news.
Every other Monday
Compare Qwen2 with
About
Qwen2 was introduced in June 2024 as Alibaba Cloud's next language-model family, spanning compact models, a mixture-of-experts option and the larger 72B tier. The release included base and instruction-tuned checkpoints. Qwen emphasized training on English, Chinese and 27 additional languages, broadening the family's multilingual ambitions beyond its original bilingual focus.
The release also targeted improvements in coding, mathematics and long-context handling. Context limits varied by checkpoint, with the larger instruction models supporting extended inputs under the documented settings. Licensing varied too: most sizes used Apache 2.0, while the 72B models used a Qwen license. Family-level comparisons therefore need the specific model size, tuning and license rather than treating every Qwen2 release as interchangeable.
Frequently asked questions
Qwen2 was released by Qwen on Thursday, Jun 6 2024.