Helper and checking models

With your own key, the small jobs around a chat run on your provider. Before, several of them quietly switched to a platform model or failed.

Which model does the helper jobsA chat starts a helper job such as a summary, a goal check, a planner or a checker. On your own key it runs on the model you set for that provider, or on a small default model from the list of your key, billed by the provider and without ZeroLog credit. On a platform model the helper stays a Featherless model.chosennot setA chat starts a helper jobsummary, /goal check,planner,checker, memory, secondopinionChat on your own keyChat on a platform modelYour setting for that providerSettings, AdvancedNo setting: a small model fromyour key's listAnthropic: Claude Haiku 5.5Runs on your keybilled by the providerno ZeroLog creditFeatherless helperas before
On your own key helper jobs stay with your provider. On the platform they stay on Featherless.
Diagram as text
  1. A chat starts a helper job, for example a summary, a /goal check, a planner or a checker.
  2. A chat on your own key uses the helper model you set for that provider under Settings, Advanced.
  3. Without a setting it uses a small model from the model list of your key. For Anthropic that is Claude Haiku 5.5.
  4. If your key lists no small model, the run model does the job itself.
  5. The job runs on your key. Your provider bills it and ZeroLog credit is not used.
  6. A chat on a platform model keeps its Featherless helper.

The checking model of an Ultracode or goal run

We tested this with a replica of the provider lists. We have not yet run every helper against a real key of each of the 25 providers, so the default for providers other than Anthropic depends on what your key lists.

See your own API key and /ultracode.