Problem feedback

After updating 1.11.7, the formula rendering on the macOS side became ugly. The device is an m4 mac mini.

When AI generates the code, if the content is long, I don't know why the generation will be interrupted, but tokens are recorded. As a result, a lot of tokens were spent but no useful information was generated.

In version 1.11.12, it is not possible to call the Modelscope community API-Inference( https://www.modelscope.cn/docs/model-service/API-Inference/intro ) service normally, and an error is prompted, with the error message being: API Error: Error from Custom: AI_TypeValidationError: Type validation failed: Value: {"id":" ","object":"chat.completion.chunk","created":1745034731,"model":"deepseek-ai/DeepSeek-R1","system_fingerprint":"","choices":[{"index":0,"delta":{"role":"","content":"","tool_calls":null,"function_calls":null,"reasoning_content":"好的"},"finish_reason":null,"message":{"role":"","content":"","tool_calls":null,"function_calls":null,"reasoning_content":""}}],"usage":{"prompt_tokens":0,"completion_tokens":0,"total_tokens":0}}. Error message: [ { "code": "invalid_union", "unionErrors": [ { "issues": [ { "received": "", "code": "invalid_enum_value", "options": [ "assistant" ], "path": [ "choices", 0, "delta", "role" ], "message": "Invalid enum value. Expected 'assistant', received ''" } ], "name": "ZodError" }, { "issues": [ { "code": "invalid_type", "expected": "object", "received": "undefined", "path": [ "error" ], "message": "Required" } ], "name": "ZodError" } ], "path": [], "message": "Invalid input" } ]

给出图片让他回答时会报错：连接 DeepSeek API 失败。这通常是由于配置错误或 DeepSeek API 账户问题。请检查您的设置并验证您的 DeepSeek API 账户状态，或购买Chatbox AI License即可立即解锁所有高级模型，无需任何配置。 Failed to deserialize the JSON body into the target type: messages[1]: data did not match any variant of untagged enum ChatCompletionRequestContent at line 1 column 175091

When selecting a model, the bottom model will be obscured and cannot be displayed in full

I've been using 3rd party API for claude and openAI models; they worked fine until 11.12, every claude model reports the following error message. OpenAi models seem to work okay, but there is still an error message generated after each response. The API works perfectly on the website version. Please fix, thanks!

Errors often occur: API Error: Error from Google Gemini: TypeError: Failed to fetch I've searched the internet and saw a related question (but not chatbox, maybe you can refer to it) https://github.com/google-gemini/deprecated-generative-ai-js/issues/29 Is this a problem with my ladder or a software issue

Android 12, version 1.11.12 – this version shows improvement over the previous one but still exhibits occasional lag.The input method automatically pops up every time I switch conversations, which worsens the lagging issue.

I'm using the Chatbox application with the Gemini 2.0 model. When the AI attempts to generate tables or structured content (such as comparing technical-economic analysis methods versus investment control methods), the application freezes and produces corrupted output - displaying numerous repetitions of the Chinese character "在" (zài) across the screen, making the response completely unusable. I've provided a screenshot as evidence showing the abnormal state when the AI tries to render tables. This issue occurs consistently whenever tables are requested, severely limiting the application's functionality. The problem does not appear with simple text responses, only when structured content like tables is generated.

On the premise that grok can be accessed, the client feedback is unable to connect to Gemini, and the model list does not have Gemini2.5Pro