Skip to main content

API Introduction

GoingMerry serves a local HTTP REST API for managing models and executing inference workloads.

Base URL​

By default, the server binds to 127.0.0.1:11434. All API requests are directed to:

http://localhost:11434

To enable network-wide access (e.g. running on a central server in your network), configure the MERRY_HOST environment variable to bind to 0.0.0.0:11434.

Conventions​

Model Names​

Model tags follow a model:tag format. The tag is optional and defaults to latest if omitted.

  • Example: gemma4:latest, llama4:8b.

Durations​

All durations (e.g. total_duration, load_duration, prompt_eval_duration, eval_duration) are returned in nanoseconds.

Streaming Responses​

For endpoints that stream output, tokens are sent as line-delimited JSON objects. You can disable streaming by passing "stream": false in the request body.