Public Inference.net model catalog source.
pull/1 fetches GET https://api.inference.net/v1/models without an API
key and stores the response in the remote cache. load/1 reads only that
cache and transforms the public fields into canonical LLMDB records.
Inference.net uses an OpenAI-compatible Chat Completions API. The catalog publishes exact model IDs, token limits, reasoning effort levels, modalities, supported endpoints, prices, and privileged-access flags.
Sources:
- https://docs.inference.net/api/api-quickstart
- https://docs.inference.net/api/coding-agents
- https://docs.inference.net/api/reasoning
- https://docs.inference.net/api/function-calling
- https://docs.inference.net/api/structured-outputs
Options: :url, :cache_dir, :req_opts, and an optional :request test
hook. The default cache directory is configured by
:inference_net_cache_dir and falls back to priv/llm_db/remote.
Summary
Functions
Transforms the public catalog into canonical LLMDB data.