BlazeRail is an OpenAI-compatible gateway. Change your base URL and every request routes to the cheapest healthy upstream serving that model. Memory, response caching and document extraction are part of the same API.
Your existing OpenAI client keeps working - same SDK, same request shape, same response shape. Point it at https://api.blazerail.com/v1.
An identical request inside the 5-minute window is served from Redis, and is not billed for tokens. Otherwise upstreams serving that model id are ordered by recorded price and measured latency. A failed call is retried on the next healthy upstream serving the same model id - never a substitute model - and the failure is recorded against the upstream that caused it.
Memory is included, retrievable by vector, full-text and knowledge graph, and scoped across sessions, keys and models. Document extraction runs on PaddleOCR and returns text plus per-page confidence.
Credits carry one platform fee. Usage then bills at the vendor’s rate, with no per-model markup. Responses served from cache are not billed for tokens.