Embeddings
Concepts and the model behind /v1/embeddings in
Embeddings.
flinq exposes a small, OpenAI-compatible HTTP API. For the full, interactive reference, with every schema, parameter and an in-browser request runner, open the live explorer.
Open the interactive API referenceThe live explorer is rendered from the same OpenAPI spec the service ships, so it is always in step with the running API.
Every request carries a Bearer key in the Authorization header:
Authorization: Bearer flq_your_key_hereKeys look like flq_... and are issued in the
flinq console. See pricing for
rates and how metering works.
Every endpoint lives under the single /v1 prefix, so
https://api.flinq.ai/v1 is the one base URL for everything: an OpenAI SDK
configured with base_url="https://api.flinq.ai/v1" works unchanged, and the
flinq-native endpoints are plain HTTP calls against the same base URL.
The surface is grouped into three categories, mirrored by the tags in the live explorer.
| Method | Path | Purpose |
|---|---|---|
POST | /v1/extract | Parse a GAEB DA XML file into its faithful bill-of-quantities tree. |
POST | /v1/chunk | Compose embed-ready chunks from a GAEB file, hierarchy-based, under your knobs. |
| Method | Path | Purpose |
|---|---|---|
POST | /v1/embeddings | Embed text into vectors. OpenAI-compatible. Supports Matryoshka dimensions. |
GET | /v1/models | List available model ids. OpenAI-compatible. |
| Method | Path | Purpose |
|---|---|---|
POST | /v1/search | Hybrid retrieval over candidate documents: dense plus BM25, fused with RRF. |
POST | /v1/rerank | Reorder a candidate pool by relevance to a query. |
GET /healthz rounds out the surface for liveness checks.
Embeddings
Concepts and the model behind /v1/embeddings in
Embeddings.
Hybrid search
How /v1/search fuses dense and lexical retrieval in
Hybrid search.
Rerank
The reranking stage in Rerank.
Try it
Send real requests in the live explorer.