Hindsight Inference API

The Hindsight Inference API serves the models behind Hindsight's own retrieval to your applications directly:

API	Endpoint	Default model	Compatible with
Embeddings	POST /v1/embeddings	hs-embed-1	OpenAI Embeddings
Reranking	POST /v1/rerank	hs-decide-1-flash	Cohere Rerank
Decision models	POST /v1/systemone	hs-decide-1-flash	TypeSafe System One

Each speaks a format you probably already use, so in most cases you switch by changing the base URL and API key in a client you already have.

• Embeddings turn text into vectors for search and clustering.
• Reranking orders candidate documents by relevance to a query.
• Decision models answer typed questions about a text: which option applies, or where on a scale it sits.

Base URL

https://inference.hindsight.vectorize.io/v1

One account, one balance

The Inference API uses the same organization, API keys and credit balance as Hindsight Cloud — there is no separate signup or wallet. A single balance covers both memory operations and inference calls. See Pricing & billing.

Getting started

1. Sign in to the Inference console with your Hindsight account.
2. Create an API key under API Keys (see Authentication).
3. Point a client at the base URL above and call one of the APIs:

curl https://inference.hindsight.vectorize.io/v1/embeddings \
  -H "Authorization: Bearer $HINDSIGHT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "hs-embed-1", "input": "Agent memory that learns."}'

What's available

The models your key can call are listed by GET /v1/models, and the console's Models page is built from it. At launch the API serves embeddings, reranking and decision models; chat/generation models stay internal to Hindsight and are not sold through the API.

API reference

The full contract — every endpoint, its request and response shapes, and the error codes — is published as an OpenAPI 3.1 document. No API key is needed to fetch it:

curl https://inference.hindsight.vectorize.io/v1/openapi.json

Point Swagger UI, Redoc, or a client generator at it to explore or integrate. It lists the same endpoints your deployment serves to API keys.

Errors

Errors return a JSON body with a detail message.

Status	Meaning	What to do
400	Invalid request (missing field, too many items, bad option)	Fix the request; detail names the field
401	Missing, invalid, revoked or expired API key	Check the Authorization header
402	Not enough credits for this request	Top up from Billing (see Pricing & billing)
403	The key has no inference access	Use a key from the Inference console's API Keys (see Authentication)
404	Unknown model	Use a model ID from GET /v1/models
429	Capacity is busy	Retry after the Retry-After header (seconds)
502	The model backend failed	Retry; contact support if it persists
503	Authentication or billing is briefly unavailable	Retry shortly