Models + runtimes.

Alluqi Bridge is the agent harness, not the underlying model. A deployment connects the models selected for the workload — cloud, private, or local.

No bundled model

Bridge ships without a model. Provider access, model licensing, and private hardware are scoped separately where needed. The specific endpoint, provider, and authentication are confirmed during assessment rather than assumed from a broad compatibility claim.

Supported runtimes

• Ollama is often the straightforward choice for local or smaller private deployments.
• vLLM is commonly better suited to a higher-throughput GPU server for team use.

Alluqi does not publish an unverified universal version list — supported versions are confirmed during configuration.

Connecting an endpoint

• Same workstation: the runtime runs on localhost, on the same machine as Bridge Desktop.
• Private network: a model server reached by private network IP and port, on an approved network path.
• Cloud: a selected cloud model provider’s API, under that provider’s terms.

Hardware reality

Private or local inference always needs a physical, GPU-capable workstation or server somewhere. Alluqi can assess the workload, size the hardware, and configure or rent the required server — see deployment options and services.