# Qwen releases Qwen2.5-VL vision-language model family

> Qwen released Qwen2.5-VL, the successor to Qwen2-VL, as a new vision-language model family. Both base and instruct models are open-sourced in three sizes: 3B, 7B, and 72B parameters. The 72B-Instruct variant is available on Qwen Chat. The release is described as a significant leap from the previous generation. Available on Hugging Face and ModelScope.

| | |
|---|---|
| **Tool** | Qwen |
| **Version** | — |
| **Kind** | announcement |
| **Published** | 2025-01-26 |
| **Observed** | 2026-08-11 |
| **Significance** | 4/5 |
| **Breaking** | no |
| **Categories** | new-model, feature |

## What changed


- New flagship vision-language model, succeeding Qwen2-VL
- Open-source base and instruct models in 3B, 7B, and 72B sizes
- 72B-Instruct available on Qwen Chat
- Available on Hugging Face and ModelScope (per platform badges in source)
- Source text is truncated; key feature details beyond vision understanding are not present


## Sources

- [blog_rss](https://qwenlm.github.io/blog/qwen2.5-vl/) — retrieved 2026-08-11


## Community

_No curated reactions recorded. Facts and community takes are kept in separate layers
and never blended._

---
Canonical: https://changelogs.info/qwen/qwen-releases-qwen2-5-vl-vision-language-model-family
Entity: https://changelogs.info/qwen
Event ID: `evt_2025-01-26_qwen_qwen2-5-vl-qwen2-5-vl-qwen2-5-vl`
Licence: event synthesis © changelogs.info, CC BY 4.0. Linked sources belong to their vendors.