Repository navigation
Can't get custom small_model to work for title generation #2640
Description
I can't get title generation working with any small_model, only without (using default claude-3.5-haiku). Help, please 🙏
Logs
Config
{
"$schema": "https://opencode.ai/config.json",
"model": "openrouter/openai/gpt-5-nano",
"small_model": "openrouter/openai/gpt-5-nano"
}Screenshot
Metadata
Metadata
Assignees
Labels
No labels
Activity
github-actions commentedon Sep 16, 2025
This issue might be a duplicate of existing issues. Please check:
chat.session.promptshould use the last used model like the opencode cli. Instead it defaults togpt-5-nano#2402:chat.session.promptshould use the last used model like the opencode cli. Instead it defaults togpt-5-nano- Related issue about model configuration not being respectedFeel free to ignore if none of these address your specific case.
rekram1-node commentedon Sep 16, 2025
I will take a look
rekram1-node commentedon Sep 17, 2025
You can get around this temporarily by using a non reasoning openai model, I will adjust the title gen logic to set reasoningEffort to low even for all gpt-5 models when generating titles
rekram1-node commentedon Sep 17, 2025
in next release we will set reasoning effort to minimal for all gpt-5 model variations when generating title and that should resolve issue for you
listenerri commentedon Mar 18, 2026
@rekram1-node
The small_model configuration still does not work for the title generation.opencode v1.2.27
Are title generation and the main conversation concurrent requests?
Originally, I set model and small_model as two local models. However, due to the VRAM limitation, only one model could be run at a time. When I changed small_model to another model, the title generation function worked well.