Key Facts
The Qwen research team officially unveiled its most powerful open-source large language model, Qwen3.5-Plus, on October 6, 2026. The news was delivered by team researcher Junyang Lin. The same day, it was covered by three sources: Reuters, 36kr, and Hacker News.
Two key aspects of the released model were highlighted. The first is its performance. Qwen3.5-Plus took first place in several benchmark leaderboards, scoring higher than Claude Sonnet 4.5 and GPT-5.1. The second is its distribution format. The model is being released as a full-parameter open-source release, meaning all of its parameters are distributed openly.
A large language model (LLM) is an artificial intelligence system trained on large collections of text to understand and generate language. Such models answer questions, write text, translate, and create code. Benchmark leaderboards are evaluation systems that measure models' performance across standard tests and are used to compare their relative standing. The model's strong showing across multiple leaderboards was noted separately by the sources.
The announcement was covered the same day by Reuters, 36kr, and Hacker News — the news appeared simultaneously in three different sources. The description "most powerful open-source model" is the Qwen team's own assessment, recorded as such in the announcement materials. ## Leading the Benchmark Leaderboards
According to October 6 reports from Reuters and 36kr, Qwen3.5-Plus leads several benchmark leaderboards. The sources confirm that the model scored higher than Claude Sonnet 4.5 and GPT-5.1 in these rankings. The same information appears in the Hacker News discussion — all three sources report the model's leaderboard standing identically.
The model tops several benchmark leaderboards, outperforming Claude Sonnet 4.5 and GPT-5.1. — Reuters, October 6, 2026
A high position on a benchmark leaderboard means the model outperforms its compared rivals on the test tasks. Qwen3.5-Plus recorded this result not in one but in several leaderboards. The two models named in the comparison — Claude Sonnet 4.5 and GPT-5.1 — are listed in the sources' materials in positions below Qwen3.5-Plus.
The quote cited in the Reuters report confirms that the model "leads specifically in several benchmark leaderboards." Benchmark leaderboards typically include tests across different domains; one model's simultaneous high ranking across several leaderboards is presented in the sources as its main achievement. All three sources — Reuters, 36kr, and Hacker News — gave mutually consistent information about the model's leaderboard results.
Qwen3.5-Plus's benchmark results, as the central news of the announcement, are listed first in every source. The information that the model scored higher than Claude Sonnet 4.5 and GPT-5.1 is presented identically in the materials of Reuters, 36kr, and Hacker News.
Full-Parameter Open-Source Release
The model is being distributed as a full-parameter open-source release. This fact was separately confirmed by the October 6 reports from Reuters and 36kr. "Full-parameter" means all of the model's parameters are presented openly — the model's internal structure is fully open.
In practice, open distribution means researchers and developers can download the model, study how it works, adapt it to their tasks, and run it independently. In such a release, the model's core components are available to the public.
That Qwen3.5-Plus is coming out specifically as a full-parameter open-source release was noted as the second major news of the announcement. Parameters are the model's internal indicators, formed during the training process and defining its behavior. The "full-parameter" format means all of these indicators are presented openly and none are kept closed. The announcement date — October 6, 2026 — is shown identically in all sources.
The information about the open release format was presented in the October 6 reports from Reuters and 36kr in a mutually confirming form. The open distribution of all of the model's parameters was noted in the announcement with special emphasis.
Coverage of the Announcement
The announcement news was delivered by Qwen research team member Junyang Lin — this name is recorded in the sources as the source of the news. 36kr and Hacker News published the story on October 6; a discussion thread on the topic was opened on Hacker News. Reuters also covered the release that day, noting the model's benchmark results and open-source format.
News of Qwen3.5-Plus appeared in all three sources on the same date — October 6, 2026. The released model is recorded as the most powerful open-source model the Qwen team has released to date.
All the key facts of the announcement — the model's name, its status as the most powerful open-source model, its leadership in benchmark leaderboards, and the full-parameter release format — were covered consistently across the three sources.
See the sources via the full Reuters report and the Hacker News discussion.
