One native CLI for images, text, video, music, speech, vision, 3D, and persistent worlds. Your weights. Your files. Nothing leaves unless you tell it to.
Real output · M4 Maxmere.run model capabilities --recommendedExplore runtimes Install a signed build from mere.run/releases, or swift build from a clone. Then:
mere.run image generate --prompt "a ceramic mug in morning light" --output mug.png
mere.run music generate "arena rock anthem, stacked vocals" --output anthem.wav
mere.run speech synthesize "Hello from mere.run" --output hello.wav
mere.run sfx generate "a ceramic mug shattering on tile" --output smash.wav
mere.run video generate "a drone flight over snowy peaks" --output flight.mp4
mere.run vision image-to-3d-trellis2 ./mug.png --output ./mug-3dEvery command uses the same binary and grammar. None requires an inference API key.
Start with mere.run model capabilities — it reads your hardware and tells you which models this machine can actually run before you spend a gigabyte finding out.
All of it ships in that one executable. This table is generated from the CLI itself, so it cannot drift from what your copy really does — and every row links to the page that owns that command.
| Command | Purpose |
|---|---|
mere.run guide | Read offline mere.run command cookbooks. |
mere.run catalog | Inspect the machine-readable command capability contract. |
mere.run image | Generate and validate image models. |
mere.run text | Run local chat, code, embedding, and anonymization workflows. |
mere.run speech | Synthesize, transcribe, diarize, and manage voice profiles. |
mere.run vision | Embed, caption, inspect, face-analyze, segment, track, pose, depth, geometry, optical flow, and OCR visual media. |
mere.run geo | Run native geospatial inference models on local Earth-observation data. |
mere.run audio | Enhance general audio locally. |
mere.run music | Generate, analyze, transcribe, and separate music locally. |
mere.run sfx | Generate sound effects locally. |
mere.run video | Generate and understand video with native Swift/MLX pipelines. |
mere.run world | Run persistent local conditioned-video world sessions. |
mere.run graph | Validate, materialize, run, and submit portable workflow graphs. |
mere.run executor | Manage local, SSH, and relay workflow executors. |
mere.run relay | Host the relay API surface directly on this machine. |
mere.run run | Inspect durable mere.run workflow reports and run directories. |
mere.run eval | Run reproducible evaluations from external, content-addressed packs. |
mere.run model | List, pull, locate, remove, inspect, optimize, and clean up models. |
mere.run adapter | List and pull verified LoRA adapters. |
mere.run status | Show local server, loaded model, and installed model status. |
mere.run gate | Run the end-to-end quality gate against installed models. |
mere.run config | Get and set persisted mere.run configuration (e.g. Hugging Face token). |
mere.run api | Serve local models through API surfaces. |
mere.run open-webui | Start the optional Open WebUI companion against a local mere.run API. |
mere.run plugin | Discover and install official mere.run companion plugins. |
mere.run setup | Choose a guided, BYOA, or manual mere.run setup path. |
mere.run agent | Install and start the optional guided local setup agent. |
Install, pull a first model, and make something in the next ten minutes.
mere.run guide works without a networkEach family is its own page: what it generates, which model to pull, and the flags that actually change the output.
Serve it, schedule it, measure it, and keep it honest.