Full Featherless catalogue and context-first search

The complete public snapshot checked on 4 October contains 51,174 unique models. ZeroLog indexes this listing locally, including models that Featherless marks as not deployed. A catalogue entry does not by itself mean that inference is available. The chat picker uses models confirmed by the authenticated API; entries without a confirmed price are not shown as free.

Models are shown by context size, largest first, then by known price. You can change the order to price, public favourites, downloads, release date, model size or name. Roleplay, Uncensored and Obliterated/Abliterated filters use Featherless's published tags. These tags and popularity figures describe the provider's catalogue; they are not ratings of ZeroLog users' activity.

The checked snapshot contains 23 models with a context of at least 262,144 tokens. Context includes input and output. A separate maximum output of 32,768 tokens does not mean that a model's total context is limited to 32K. A plan's context allowance also does not enlarge an individual model's context.

Public metadata is refreshed in the background. Incomplete responses and large unexpected losses retain the previous index, and provider retry delays are respected. Search uses the local index with bounded result pages. Existing model permissions and the customer pause remain in effect.

Recommendations remain separate from the full listing and the platform-admin model section. They require documented tests; public favourites and downloads do not substitute for those tests.

Open the model catalogue | Featherless model API documentation