当具备思考能力的ai模型的思考过程正在进行时，可能会出现思考失败（即思考流程被强行终止）的情况，随之该消息会被自动撤回，请求作者查明原因。

[player532]

这个思考过程进行失败的出现频率是非常不稳定的，一旦出现思考失败的情况，只要不编辑提问内容，便很有可能会一直触发此问题，但问题的触发根本原因尚未知晓。

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

Title: When the thinking process of a specific AI model is in progress, there is a high probability that the thinking may fail, and then the message will be automatically withdrawn, requesting the author to find out the reason.

[player532]

如果作者没有完全理解我的本意，请查看下面的GIF，下面的GIF已经复现了该问题的描述，经测试，长对话也可能会导致问题的复现，但问题复现频率尚未知晓。

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

If the author does not fully understand my intention, please view the GIF below. The GIF below has reproduced the description of the problem. After testing, long conversations may also cause the problem to recur, but the frequency of the problem recurrence is not yet known.

[player532]

这是在问题出现之前，模型想法内部的大致内容，疑似思考过程被完全略过了，而正式输出的结果误入进了模型想法当中，才导致出现了问题，但在我看来，很有可能不仅仅是因为这个原因。

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

This is roughly what is inside the model idea before the problem arises

[Amery2010]

According to the official documentation
https://ai.google.dev/gemini-api/docs/thinking-mode#limitations
Text and image input only, No audio support

[player532]

According to the official documentation https://ai.google.dev/gemini-api/docs/thinking-mode#limitations Text and image input only, No audio support

@Amery2010 作者，这个真的与音频输入的问题无关，因为我已经测试过了，即使在单个话题当中全程仅输入图片与文字，但思考流程被强行终止的问题却再次复现了：

Screenrecording_20241222_202639.mp4

[player532]

@Amery2010 这张图片将作为长截图的证据，我是不会无缘无故欺骗你的：

[Issues-translate-bot]

Bot detected the issue body's language is not English, translate it automatically.

---

@Amery2010 This picture will serve as proof of the long screenshot, I won’t lie to you for no reason:

[Amery2010]

OK, I will track this problem.

1. The new model itself may have errors.
2. The code generated errors during the generation process.
3. There are problems with the logic of the code I refactored.

This problem is difficult to solve in a short time, you can consider using other stable models first.

[player532]

OK, I will track this problem.

1. The new model itself may have errors.
2. The code generated errors during the generation process.
3. There are problems with the logic of the code I refactored.

This problem is difficult to solve in a short time, you can consider using other stable models first.

在我这里又出现了一个思考时的错误，在这里出现了自动中断的错误，导致无法继续输出。作者可以看一下可能是这个问题吗？

[Amery2010]

Ok, I'll try to fix the problem

[player532]

OK, I will track this problem.

1. The new model itself may have errors.
2. The code generated errors during the generation process.
3. There are problems with the logic of the code I refactored.

This problem is difficult to solve in a short time, you can consider using other stable models first.

@Amery2010 我想我可能找到根本原因了，它基本上是在长对话当中出现此问题的，也就是多轮对话，当单话题对话轮次越多，该issues所描述的bug就会更容易的复现出来，作者也可以亲自测试。

[player532]

OK, I will track this problem.

1. The new model itself may have errors.
2. The code generated errors during the generation process.
3. There are problems with the logic of the code I refactored.

This problem is difficult to solve in a short time, you can consider using other stable models first.

在我这里又出现了一个思考时的错误，在这里出现了自动中断的错误，导致无法继续输出。作者可以看一下可能是这个问题吗？

Ok, I'll try to fix the problem

@Amery2010 关于我上次向你所描述的问题，那种可能属于另一种错误，本质上模型思考过程中自动出现终止并撤回的现象并未得到修复。所以劳烦您了。

[player532]

@Amery2010 在这里，模型的最终答案输出被错误的写入在了模型思考过程当中，思考过程也被错误的无视掉了，作者可以很直观的看到图片中的现象