Qwen2.5-14B-Instruct

weights

open(Apache-2.0, on HF)

data

closed

code

partial(inference/chat template

license

Apache-2.0(OSI permissive)

Apache-2.0 weights anyone can download, but the post-training data and the training code are closed, so this is open weights rather than open source.

• https://huggingface.co/Qwen/Qwen2.5-14B-Instruct recorded 2026-08-13

  `license:apache-2.0` in the repo tags and `"gated":false` in the embedded repo state, so the 14.7B BF16 weights download without a barrier; the card carries a quickstart and chat template only, with no post-training pipeline or SFT/RLHF mixture

The single declared SKU, the 14B-Instruct model, reads 3,137,049 Hugging Face downloads in the trailing 30 days, which lands in the 1M-10M band at level 4. It is one of the most-pulled mid-size instruct models in the widely used Qwen2.5 family.

• https://huggingface.co/Qwen/Qwen2.5-14B-Instruct recorded 2026-08-13

  3,137,049 downloads in the trailing 30 days

Capable mid-size (14B) 2024 general chat model; strong-for-size but well below 2026 frontier and below larger Qwen2.5/3 SKUs. HF card does not inline MMLU/MMLU-Pro numbers; scored 3 on size/class with medium confidence.

• https://huggingface.co/Qwen/Qwen2.5-14B-Instruct recorded 2026-08-13

  14B instruct general-chat model card, vendor capability claims; defers benchmarks to Qwen2.5 blog