[Feature Request]: Bring Back Skip Button or Adjustable Explicit Content in the Next Upcoming Version of GLM

提交前确认 · Pre-submission checklist

• 我已搜索过现有 issue，确认这不是重复提议 / I searched existing issues and confirmed this isn't a duplicate.
• 我已阅读 CONTRIBUTING.md / I've read CONTRIBUTING.md.

问题类别 · Category

UI / 界面体验 · UI / UX

涉及的 Agent 框架 · Agent framework

ZCode Agent（自研）

使用场景 · Use case

It is concerning to observe that the latest version of GLM has demonstrated performance degradation, resulting in extended output generation times. Extended chat inputs now require approximately one minute, with response times frequently reaching two to three minutes, which represents a notable efficiency deficit compared to Deepseek, particularly without an accessible skip function. It is noteworthy that GLM 5.3 has exhibited decreased processing speed and reduced computational capability relative to its predecessor versions. The removal of the previously available 'Skip' button constitutes a significant functional limitation and negatively impacts user productivity. The objective of implementing roleplay fanfiction long conversation input chat sessions is to generate outputs expeditiously and effectively; however, the absence of skip functionality in version 5.3 results in substantially prolonged response times, frequently requiring two to three minutes to produce outputs. This situation not only creates operational inefficiency but also substantially diminishes the overall user experience.When Using GLM 5.3 and GLM 5.2 exhibit distinct behavior when it comes to the "Stop Thinking" skip button. In the first video, when GLM 5.2 is used for a brief 13-second output, a skip button is present between 0:00:07 and 0:00:15. However, in the second video, when GLM 5.3 is utilized for a longer AI output on the chat.z.ai web UI, there is no skip button present between 0:01:10 and 0:01:16, resulting in a slower response compared to GLM 5.2.The utilization of GLM 5.3 and GLM 5.2 yields dissimilar outcomes in relation to the "Stop Thinking" skip button. In the initial video, when GLM 5.2 is implemented for a duration of 13 seconds, a skip button is visible between 0:00:07 and 0:00:15. Conversely, in the subsequent video, when GLM 5.3 is engaged for an extended AI output on the chat.z.ai web UI, a skip button is absent between 0:01:10 and 0:01:16, consequently leading to a delayed response compared to GLM 5.2. Both GLM 5.3 and Flash exhibit disastrous and sluggish performance in generating responses to long-length inputs on the chat.z.ai website. formal report detailing the technical issues identified in GLM 5.3. The content moderation guidelines implemented in GLM 5.3 are substantially more stringent than those in GLM 5.2. I respectfully request that this report be considered for inclusion in the development roadmap for the December 2026 GLM release. I recommend reconsidering the guideline parameters to restore the content flexibility observed in GLM 5.2, particularly regarding explicit content in fanfiction roleplay contexts, for subsequent versions including GLM 5.5 and GLM 6.0. Furthermore, I recommend a comprehensive review of the comparative analysis materials examining the explicit content policies between GLM 5.2 and GLM 5.3. I request that these findings be communicated to the development team to inform future iterations of GLM 5.5 and GLM 6.0. Historical data indicates that GLM 5.2 accommodated certain minor explicit content parameters, which have been eliminated in version 5.3. Additionally, the current GLM 5.3 guidelines restrict my ability to generate fanfiction roleplay content, as the updated version prioritizes cybersecurity considerations over fanfiction roleplay functionality. Consequently, I propose that the development team evaluate an equilibrium between cybersecurity requirements and adjusted content guidelines for explicit material, consistent with the consistent with the parameters established in GLM 5.2.formal report detailing the technical issues identified in GLM 5.3. The content moderation guidelines implemented in GLM 5.3 are substantially more stringent than those in GLM 5.2. I respectfully request that this report be considered for inclusion in the development roadmap for the December 2026 GLM release. I recommend reconsidering the guideline parameters to restore the content flexibility observed in GLM 5.2, particularly regarding explicit content in fanfiction roleplay contexts, for subsequent versions including GLM 5.5 and GLM 6.0. Furthermore, I recommend a comprehensive review of the comparative analysis materials examining the explicit content policies between GLM 5.2 and GLM 5.3. I request that these findings be communicated to the development team to inform future iterations of GLM 5.5 and GLM 6.0. Historical data indicates that GLM 5.2 accommodated certain minor explicit content parameters, which have been eliminated in version 5.3. Additionally, the current GLM 5.3 guidelines restrict my ability to generate fanfiction roleplay content, as the updated version prioritizes cybersecurity considerations over fanfiction roleplay functionality. Consequently, I propose that the development team evaluate an equilibrium between cybersecurity requirements and adjusted content guidelines for explicit material, consistent with the parameters established in GLM 5.2.

建议方案 · Proposal

On the website chat.z.ai, GLM 5.5 is intended to function similarly to GLM 5.2 while incorporating combined features from both GLM 5.3 and GLM 5.2. The relevant distinction pertains to the Stop Thinking Button feature and GLM 5.2's handling of explicit content in fanfiction roleplay scenarios. GLM 5.5 is positioned as an enhanced model variant. Additionally, GLM 5.5 should not produce slow output generation that extends for several minutes when processing extended user input text. is anticipated that the forthcoming iteration of GLM will integrate the capabilities of both GLM 5.2 and GLM 5.3, while maintaining expedited output generation in the web user interface, rather than exhibiting the slower response characteristics observed in GLM 5.3, and will implement content moderation policies consistent with those established in GLM 5.2.

预期价值 · Expected value

No response

你认为的优先级 · Your perceived priority

高 · High

你使用的 ZCode 版本 / 环境 · ZCode version / environment

chat.z.ai Website UI Interface

补充材料 · Additional context

Comparative Analysis: Video Evidence SlowDemonstrating the Processing Speed Explicit Content Handling Capabilities of GLM 5.3 and Video Comparison of GLM 5.3 and 5.2 https://github.com/user-attachments/assets/c019e7d0-add0-4c6c-a781-9e81a514e04c
https://github.com/user-attachments/assets/6c2a59e6-90af-47c6-a1e4-fc65d5fe7f0e
https://github.com/user-attachments/assets/ddff05ad-d941-4eea-b376-c954e1d06a0a
https://github.com/user-attachments/assets/90cb5815-aa33-4f01-a468-d93dcc16d9f6

https://github.com/user-attachments/assets/6a4b32aa-e843-4f81-ad74-fc1b86448e16

The report concerns the chat.z.ai Website UI, GLM 5.3/5.2 behavior, and attached comparison videos; no repository file, test, or code entry point is identified. Start by reproducing the missing Stop Thinking button and slow long-input responses in that UI; completion criteria are not specified beyond changing these behaviors and reviewing content-moderation differences.