API REFERENCE
OpenAI-compatible inference.
STABLE INTERFACE · 2 MODELS AVAILABLE
Base URL
https://api.vorth.com/v1Authentication
Send the dedicated marketplace credential as a bearer token. Never place credentials in query strings or client-side applications.
Authorization: Bearer YOUR_API_KEYEndpoints
GET /modelsAvailable models, flavors, and capabilitiesPOST /chat/completionsStreaming and non-streaming chat inferenceModel catalog
slick-qwen3.6-35b-a3b · Slick Qwen3.6 35B-A3BREADY · release inkling-qwen3.6-v1.0.2-vorth.1 · coverage: doom-loop repair, thinking-only exhaustion repair · pricing: unpublishedsharp-qwen3.6-35b-a3b · Sharp Qwen3.6 35B-A3BREADY · release inkling-qwen3.6-v1.0.2-vorth.1 · coverage: doom-loop repair, thinking-only exhaustion protection · pricing: unpublishedModel identifiers, tiers, releases, readiness, capabilities, and pricing state are published by GET /v1/models. Protection coverage is a release fact published by a separate Vorth-generated manifest; where that manifest is missing or stale this page renders coverage as not published rather than inferring it.
Compatibility
Ready releases use the OpenAI chat-completions shape. Unsupported parameters fail explicitly rather than being silently ignored. Model identifiers, context limits, prices, and supported features are returned by /models.
Data use
Complete prompts, visible responses, raw thinking, and request-linked token sequences are not retained in Vorth’s application dataset. Where data collection is permitted, Vorth may retain sufficiently populated aggregate measurements from the private thinking phase to evaluate service quality and efficiency. Review the privacy notice before routing traffic.
Limits and support
Traffic is subject to request, spend, concurrency, and duration limits. The service may reject or terminate requests to preserve reliability or enforce the acceptable-use policy. Contact contact@vorth.com.