Qwen3.5-flash LLM Provider Error Appears Even When a Different Model Is Selected

Developers
1

Issue Description

When using the chat feature in Anna Dashboard, I sent a normal message and received the following error message in Chinese:

未找到支持模型 qwen3.5-flash 的 LLM 提供商

img_v3_02130_139003e0-7b3a-4a66-ad80-288fb496411g
img_v3_02130_139003e0-7b3a-4a66-ad80-288fb496411g1460×768 113 KB

img_v3_02130_8b567410-4e8a-4979-bd5a-11b3f57a307g
img_v3_02130_8b567410-4e8a-4979-bd5a-11b3f57a307g1379×742 110 KB

This means that no LLM provider was found for the model qwen3.5-flash.

However, I did not actively select qwen3.5-flash. As shown in the screenshot, the model selector at the bottom of the chat UI displays Qwen3.7 Max, and the current session is using Zen Mode with 1 Agent. Therefore, the error appears to indicate that the system is still trying to call qwen3.5-flash internally, or that the session configuration, model routing, default model fallback, and frontend model selection are not consistent.

Account Information

  • Account name: orailkane
  • Account email: orailkane@gmail.com
  • Page path: anna.partners/dashboard

Steps to Reproduce

  1. Log in to the Anna account listed above.
  2. Open Anna Dashboard.
  3. In the main chat interface, use the currently visible model configuration. In the screenshot, it is shown as Qwen3.7 Max.
  4. Keep Zen Mode and 1 Agent enabled.
  5. Send a normal message, for example: Hi.
  6. The page returns the error: 未找到支持模型 qwen3.5-flash 的 LLM 提供商.

Actual Result

The chat message does not complete successfully, and the page shows the following error:

未找到支持模型 qwen3.5-flash 的 LLM 提供商

The model name in the error message does not match the model currently shown in the UI.

Expected Result

  • If I selected Qwen3.7 Max, the system should route the request to the provider associated with that model.
  • If the system needs to fall back to another model, the fallback model should have an available LLM provider.
  • If a model is unavailable, the error message should clearly state the actual model being requested, why fallback was triggered, and what the user can do to recover or select an available model.
  • The model displayed in the frontend selector should stay consistent with the model actually used by the backend request.

Impact

This issue makes it difficult to determine whether the selected model is unavailable or whether the system is internally calling a different model by mistake. For developers or frequent users, this affects debugging and reduces trust in the model selector and session configuration state.

Suggested Areas to Investigate

  • Check whether this account has a saved session configuration that still references qwen3.5-flash as the default model, agent model, or fallback model.
  • Check whether Zen Mode or 1 Agent has a separate model routing configuration independent of the frontend model selector.
  • Check whether the value displayed in the frontend model selector matches the actual model field in the backend request payload.
  • Check whether the provider registry is missing a provider mapping for qwen3.5-flash, or whether this model has been deprecated but is still referenced by configuration.

Attachment: the issue screenshot is included with this feedback.