# `LLMDB.Sources.Superlinked`
[🔗](https://github.com/agentjido/llmdb/blob/main/lib/llm_db/sources/superlinked.ex#L1)

Cached source for the Superlinked Inference Engine model catalog.

`pull/1` requests `GET https://api.superlinked.com/v1/models` with a bearer
token. Set `SIE_API_KEY` or `SUPERLINKED_API_KEY`, or pass `:api_key`.
Tests can pass `:req_opts`, `:cache_dir`, and `:url`.

`load/1` reads the cache without a network request or an API key. The source
keeps models from an earlier successful pull when an account-specific reply
omits them.

Token-output models use the OpenAI-compatible Chat Completions contract.
Dense-output models use the OpenAI-compatible Embeddings contract. Score
output records rerank capability, but stays catalog-only unless the model also
has an executable chat or dense embedding contract. Native sparse,
multivector, JSON extraction, and OCR operations are catalog metadata only.

# `transform`

Transforms the Superlinked catalog into canonical LLMDB data.

---

*Consult [api-reference.md](api-reference.md) for complete listing*
