You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[Bug] Background jobs and subagents force GPT-5 Nano even when small_model is set to gpt-5.4 and opencode provider is disabled #24812
Trigger any action that spawns a subagent (especially via oh-my-openagent — Explore, analyze, etc.).
Observe that subagent/background task still runs on GPT-5 Nano.
Expected behavior
All agents, subagents, and background/lightweight tasks should respect the configured model / small_model and never fall back to GPT-5 Nano when that provider is disabled and a different model is explicitly set.
Actual behavior
Subagents (particularly those started by OMO_INTERNAL_INITIATOR) ignore the config and default to GPT-5 Nano.
Environment
OpenCode version: lastest version
OS: Windows
Plugin: oh-my-openagent, true-mem
Provider: OpenAI
Additional context
This seems to be a regression or hard-coded fallback inside the subagent / internal initiator logic, bypassing user configuration for small_model and disabled providers.
It forces usage of a model I explicitly do not want (and have disabled the provider for).
Activity
github-actions commentedon Apr 28, 2026
This issue might be a duplicate of existing issues. Please check:
explore)If your issue is distinct from these (e.g., specific to the
oh-my-openagentOMO_INTERNAL_INITIATORpath), please add a note clarifying how it differs.