Qwen3.5 is the large language model series developed by Qwen team, Alibaba Cloud.
active 2026-02-16 → 2026-04-16 (UTC)
Activity over time
Daily event counts in the loaded window
Line chart, 60 days from 2026-02-16 to 2026-04-16. Pushes: 3 total, peak 1 in a day. Pull requests: 1 total, peak 1 in a day. Issues: 26 total, peak 3 in a day. Comments: 51 total, peak 5 in a day. Stars: 644 total, peak 90 in a day.
- Pushes
- Pull requests
- Issues
- Comments
- Stars
Stars, PRs, issues and forks are under-captured in the later part of this window. GH Archive progressively stopped capturing non-push events during 2026 — −95% or worse by the end of the window. Every series here except Pushes fades for that reason, so a decline above reflects the archive, not this repository. Pushes stay reliable throughout, so read them, and the contributor counts derived from them, as the real signal. Data health has the measurements.
Top contributors
Pushes, PRs, issues, reviews and comments — stars and forks excluded, so this is contribution rather than popularity
| Contributor | Contributions | Pushes | PRs | Comments |
|---|---|---|---|---|
| jklj077 | 12 | 3 | 1 | 2 |
| github-actions[bot] | 10 | 0 | 0 | 6 |
| JJJYmmm | 6 | 0 | 0 | 6 |
| roj234 | 6 | 0 | 0 | 6 |
| lucasjinreal | 4 | 0 | 0 | 2 |
| Ossamanal | 3 | 0 | 0 | 1 |
| wenmengzhou | 2 | 0 | 0 | 2 |
| jpg1024 | 2 | 0 | 0 | 2 |
| FactrueSolin | 2 | 0 | 0 | 1 |
| 948024326 | 1 | 0 | 0 | 0 |
| ToCaGlint | 1 | 0 | 0 | 1 |
| akawincent | 1 | 0 | 0 | 1 |
| DragonAura | 1 | 0 | 0 | 0 |
| ouroborosscr | 1 | 0 | 0 | 0 |
| LIUKAI0815 | 1 | 0 | 0 | 1 |
| catchy666 | 1 | 0 | 0 | 1 |
| LianShuQuan | 1 | 0 | 0 | 1 |
| latent-variable | 1 | 0 | 0 | 1 |
| sky-fly97 | 1 | 0 | 0 | 1 |
| CloudyDory | 1 | 0 | 0 | 0 |
Recent activity
Latest issues, pull requests and releases
- Issue comment#35ToCaGlint2026-04-16 01:52Qwen3.5-series models always think in english
- Issue#134jklj0772026-04-14 11:38incorrect processing of a line of code
- Issue#65github-actions[bot]2026-04-12 08:34MTP finetune support
- Issue#80github-actions[bot]2026-04-12 08:34微调代码什么发布，基于几个框架微调都报各种错误，QA和VQA的微调代码，sft和grpo
- Issue#33github-actions[bot]2026-04-12 08:34Vocabulary issue with undertrained long tokens causing semantic collapse and hallucinations / 词表中有未被充分训练的词汇，模型在这些词汇上表现很差
- Issue comment#80github-actions[bot]2026-04-12 08:34微调代码什么发布，基于几个框架微调都报各种错误，QA和VQA的微调代码，sft和grpo
- Issue comment#131EBOC8342026-04-11 16:25chat template emits empty historical <think> blocks, causing prompt drift and cache invalidation
- Issue#77github-actions[bot]2026-04-11 08:28Qwen3.5 plus sometimes outputs "(Earlier text omitted)" and starts reasoning from the middle
- Issue comment#77github-actions[bot]2026-04-11 08:28Qwen3.5 plus sometimes outputs "(Earlier text omitted)" and starts reasoning from the middle
- Issue comment#78github-actions[bot]2026-04-11 08:28包含中文数字文件名，模型会自己添加空格
- Issue comment#131latent-variable2026-04-11 07:42chat template emits empty historical <think> blocks, causing prompt drift and cache invalidation
- Issue comment#35rxy02102026-04-10 01:44Qwen3.5-series models always think in english
- Issue comment#35lcy4122026-04-09 04:50Qwen3.5-series models always think in english
- Issue comment#122jklj0772026-04-07 11:50使用最新的VLLM部署qwen3.5-9B，输出的内容都在think中，lmdeploy也是。
- Issue#125jklj0772026-04-07 11:03Qwen3.5-9B / 35B-A3B produce XML-style tool calls instead of JSON under SGLang tool-call parser
- Issue comment#122LIUKAI08152026-04-07 06:50使用最新的VLLM部署qwen3.5-9B，输出的内容都在think中，lmdeploy也是。
- Issue#126CloudyDory2026-04-06 14:20Unable to reproduce HLE benchmark score with tool-use (Qwen3.5-122B-A10B)
- Issue comment#35jpg10242026-04-05 10:34Qwen3.5-series models always think in english
- Issue comment#33github-actions[bot]2026-04-05 08:29Vocabulary issue with undertrained long tokens causing semantic collapse and hallucinations / 词表中有未被充分训练的词汇，模型在这些词汇上表现很差
- Issue comment#7github-actions[bot]2026-04-05 08:29[Qwen3.5-397B-A17B] Incomplete JSON output when "Thinking" is disabled (via OpenRouter)
- Issue comment#35jpg10242026-04-05 08:27Qwen3.5-series models always think in english
- Issue comment#122jdnt2026-04-05 02:58使用最新的VLLM部署qwen3.5-9B，输出的内容都在think中，lmdeploy也是。
- Issue comment#22wqh171012026-04-03 07:57Unexpected spacing between Chinese and English characters in mixed output
- Issue comment#100jklj0772026-04-02 07:12[Bug] Chat interface corrupts Python code by injecting extra spaces before curly braces in f-strings
- Issue comment#110LianShuQuan2026-04-02 04:56Unable to reproduce SWE-Bench Verified scores due to missing agent framework and max_turns configuration
Totals cover only the window loaded into ClickHouse and count events, not GitHub's lifetime totals — 644 stars here means stars gained during the window, not the repo's star count.