# Phoeniqs AI Model Service

Phoeniqs Model as a Service (MaaS) gives your organisation API access to a curated catalogue of open-source large language models: chat, reasoning, embedding, reranking, multimodal, OCR, and speech-to-text.

Access them through a single OpenAI-compatible endpoint at https://maas.phoeniqs.com/. Models are hosted entirely on Swiss sovereign infrastructure, served with vLLM for optimised inference, and billed per token through your subscription's monthly credit allowance.

You can use MaaS to power assistants, agentic coding workflows, document processing, semantic search, RAG pipelines, and integrations with tools such as Cursor and OpenCode.

One subscription supports multiple API keys. All keys on the same subscription share the same credit pool.


# Model naming

Model identifiers on the Phoeniqs Model Service follow semantic versioning at the major version only. The name you use in API calls stays stable across minor and patch releases, for example inference-glm5 (GLM 5), not GLM 5.1 or GLM 5.2. See the Active Models table for the current identifier for each model.

This means you do not need to actively update the model name in your applications or agents at each minor or patch release, your existing API calls keep working against the same identifier while Phoeniqs upgrades the model behind it.

When we ship a minor or patch update behind an existing model ID, we do not send a separate email announcement; the change is recorded in the Changelog. We do notify customers by email when:

  • a new major version is promoted to the active catalog, or
  • a previously reachable model ID is permanently removed, even if the underlying release was only a minor or patch change.

For those cases, you may need to update your integration, for example, when migrating to a new major version or before a decommissioned model ID stops responding.


# Quick links

Rate limits, tokens-per-minute ceilings, and throughput per subscription tier.
rate-limits-and-throughput/
View the full list of models currently live and available for inference.
active-models/
Check model health and availability on the status dashboard.
https://status.kvant.cloud/
Step-by-step guides for inference, API calls, subscriptions, and more.
model-service-guides/
Connect Phoeniqs MaaS to your existing applications and platforms.
model-service-integrations/
Subscription plans, credits, and top-ups.
subscriptions/