Description:
I am encountering a persistent issue where the plugin stops using my configured relay/proxy model and unexpectedly switches to claude-opus-4-6.
Observed Behavior:
Trigger: This happens primarily when the context length is very large or when a request takes a long time to process (likely a timeout issue).
Unexpected Fallback: Instead of respecting my custom model settings (GLM-5 via relay), the plugin silently falls back to claude-opus-4-6.
Infinite Loop: Once the fallback occurs, the plugin enters an infinite loop of requests to claude-opus-4-6. As shown in the attached logs, these calls frequently result in 0-token usage, yet they keep firing repeatedly.
Steps to Reproduce:
Configure the plugin to use a custom relay model (e.g., GLM-5).
Execute a command that triggers a large context or requires significant processing time.
Observe the logs: the plugin switches to claude-opus-4-6 and initiates an endless loop of requests.
Environment:
Project: claude-code-chat
Configured Model: GLM-5 (via relay)
I have attached a screenshot of the logs showing the constant, zero-token calls to claude-opus-4-6.
Description:
I am encountering a persistent issue where the plugin stops using my configured relay/proxy model and unexpectedly switches to claude-opus-4-6.
Observed Behavior:
Trigger: This happens primarily when the context length is very large or when a request takes a long time to process (likely a timeout issue).
Unexpected Fallback: Instead of respecting my custom model settings (GLM-5 via relay), the plugin silently falls back to claude-opus-4-6.
Infinite Loop: Once the fallback occurs, the plugin enters an infinite loop of requests to claude-opus-4-6. As shown in the attached logs, these calls frequently result in 0-token usage, yet they keep firing repeatedly.
Steps to Reproduce:
Configure the plugin to use a custom relay model (e.g., GLM-5).
Execute a command that triggers a large context or requires significant processing time.
Observe the logs: the plugin switches to claude-opus-4-6 and initiates an endless loop of requests.
Environment:
Project: claude-code-chat
Configured Model: GLM-5 (via relay)
I have attached a screenshot of the logs showing the constant, zero-token calls to claude-opus-4-6.