The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.
IT Home[4] reports that on the evening of 14 August 2026 the "Qwen" WeChat account announced Qwen3.8-27B "正式开源" (officially open-sourced); Kingy[5] AI times the Hugging Face release at 15:00 UTC on 14 August, and OpenRouter[6] lists it as released Aug 14, 2026.
Alibaba Cloud's repository for the model: a native multimodal dense open-weight model under Apache 2.0, 27B parameters, 262K native context extendable to 1M, said to outperform Qwen3.7-Plus overall in software engineering and office workflows.[1]
Places Qwen3.8-27B (Apache 2.0, 14 Aug 2026) after the 3 Aug Qwen3.8-Max cloud release and the 12 Aug open weights of the 2.4T-A95B Max model, which use a custom licence.[1]
IT Home, 14 Aug 2026: on the evening of the 14th Alibaba's Qwen account announced Qwen3.8-27B as open source for all developers; a native multimodal dense model, 262K native context extendable to 1M tokens with YaRN, a new reasoning_effort control, and performance above Qwen3.7-Plus in coding and office scenarios.
Kingy AI's launch-day review: released at 15:00 UTC on 14 August; the checkpoint has 27.78 billion parameters and is a 55.6 GB download; every launch score comes from Qwen and no independent reproduction existed at launch; managed Qwen Cloud endpoint "coming soon".
Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most. A picture that shows something else, or shows it badly, is looked at too, and moved down or replaced.