All corrections
Wikipedia June 18, 2026 at 06:27 AM

en.wikipedia.org/wiki/Qwen

3 corrections found

1
Claim
In January 2025, Qwen2.5-VL was released with variants of 3, 7, 32, and 72 billion parameters.
Correction

This folds a later 32B release into the initial January 2025 launch. Alibaba’s January 26, 2025 Qwen2.5-VL announcement listed only 3B, 7B, and 72B models; the 32B model was announced separately on March 24, 2025.

Full reasoning

Alibaba’s own release posts contradict this timeline.

  • In the official January 26, 2025 Qwen2.5-VL launch post, Qwen says it opened the models in three sizes: 3B, 7B, and 72B.
  • In a separate official post dated March 24, 2025, Qwen introduced Qwen2.5-VL-32B-Instruct and described it as a new 32B model built on the earlier Qwen2.5-VL series.

So the article’s sentence is inaccurate because it treats 32B as part of the original January 2025 release, when Alibaba announced that size later in March 2025.

2 sources
2
Claim
All models except the 72B variant are licensed under the Apache 2.0 license.
Correction

This overstates the Apache licensing. The 3B Qwen2.5-VL model was released under the Qwen Research License, not Apache 2.0.

Full reasoning

The statement says every Qwen2.5-VL model except 72B is Apache 2.0, but Alibaba’s 3B model contradicts that.

The official Hugging Face license file for Qwen2.5-VL-3B-Instruct is titled “Qwen RESEARCH LICENSE AGREEMENT”, not Apache 2.0. That means at least one non-72B Qwen2.5-VL model was not Apache-licensed, so the sentence as written is false.

2 sources
3
Claim
Qwen2-VL December 2024
Correction

This release date is wrong. Alibaba’s official Qwen2-VL announcement was on August 29, 2024, not December 2024.

Full reasoning

Alibaba’s official Qwen2-VL launch post is dated August 29, 2024 and says, “today we are thrilled to release Qwen2-VL.” In that same announcement, Alibaba says it open-sourced Qwen2-VL-2B and Qwen2-VL-7B and released the API for Qwen2-VL-72B.

So labeling the Qwen2-VL family in the model table as a December 2024 release is inconsistent with the official release announcement, which places the family launch in August 2024.

1 source
Model: OPENAI_GPT_5 Prompt: v1.16.0