Problem (one or two sentences)
The context window token calculation intermittently bugs out and jumps up to 1M (1,000,000) tokens, ignoring the configured context limit (e.g., 250k tokens) set in the UI.
Context (who is affected and when)
Users working on tasks with auto-condensing or dynamic context updating, where the token count bar unexpectedly spikes/resets to a 1M token ceiling.
Reproduction steps
Set up a task with a restricted context window (e.g., 250k token limit).
Run requests or let the context auto-condense/update.
Observe the token usage bar: the context capacity/counter randomly jumps up to 1M tokens instead of staying within the designated limit.
Expected result
The context token bar and limit should accurately reflect and strictly adhere to the configured model/session bounds (e.g., max 250k tokens).
Actual result
The context token count or ceiling suddenly jumps to 1M tokens (1,000,000 tokens) during context condensing/refreshing.
Variations tried (optional)
Happens intermittently across different prompts and conversation lengths.
App Version
Version: 3.74.0 (6c23a1c)
API Provider (optional)
None
Model Used (optional)
No response
Zoo Code Task Links (optional)
No response
Relevant logs or errors (optional)
Problem (one or two sentences)
The context window token calculation intermittently bugs out and jumps up to 1M (1,000,000) tokens, ignoring the configured context limit (e.g., 250k tokens) set in the UI.
Context (who is affected and when)
Users working on tasks with auto-condensing or dynamic context updating, where the token count bar unexpectedly spikes/resets to a 1M token ceiling.
Reproduction steps
Set up a task with a restricted context window (e.g., 250k token limit).
Run requests or let the context auto-condense/update.
Observe the token usage bar: the context capacity/counter randomly jumps up to 1M tokens instead of staying within the designated limit.
Expected result
The context token bar and limit should accurately reflect and strictly adhere to the configured model/session bounds (e.g., max 250k tokens).
Actual result
The context token count or ceiling suddenly jumps to 1M tokens (1,000,000 tokens) during context condensing/refreshing.
Variations tried (optional)
Happens intermittently across different prompts and conversation lengths.
App Version
Version: 3.74.0 (6c23a1c)
API Provider (optional)
None
Model Used (optional)
No response
Zoo Code Task Links (optional)
No response
Relevant logs or errors (optional)