ColdState API
Online
Cut your LLM
retrieval costs.
Keep the accuracy.
ColdState replaces embedding-based search with deterministic, whole-word semantic retrieval. No GPU inference. No vector databases. No hallucinated results. Just fast, explainable search that your AI agents can trust.
2.8M
Vocabulary
0
Inference
35
Domains
100%
Deterministic
$ Cost comparison per 1M queries
Embedding generation
$300
$0
-100%
Vector DB hosting
$200/mo
$0
-100%
Search queries
$300
$20
-93%
GPU compute
$500+/mo
$0
-100%
/ API Endpoints
POST
/v1/search
Search by collection
POST
/v1/indexes/:id/search
Search by index
POST
/v1/indexes/:id/explain
Score explanation
POST
/v1/indexes
Create index
GET
/v1/indexes
List indexes
GET
/v1/indexes/:id/documents
Browse documents
GET
/v1/usage
Usage metrics
Checking API status...