List Models
mixtrain model listList all models in the workspace.
Options:
| Option | Description |
|---|---|
--provider, -p | Filter by provider: openai, anthropic, workspace, etc. |
mixtrain model list # Workspace models
mixtrain model list --provider openai # OpenAI models
mixtrain model list -p workspace # All workspace modelsDownload Model
mixtrain model download <name>Download model source files to a local directory for editing.
Options:
| Option | Description |
|---|---|
--dir, -d | Target directory (defaults to ./<name>/) |
--force, -f | Overwrite existing directory |
# Download to default directory (./<name>/)
mixtrain model download my-model
# Download to custom directory
mixtrain model download my-model --dir ./local-copy/
# Overwrite existing directory
mixtrain model download my-model --forceUpload Model
mixtrain model upload <name> --dir <path>Upload local model files to update the model on the platform.
Options:
| Option | Description |
|---|---|
--dir, -d | Local directory containing model files (required) |
# Upload all files from directory
mixtrain model upload my-model --dir ./local-copy/
# Typical workflow: download → edit → upload
mixtrain model download my-model
# ... edit files locally ...
mixtrain model upload my-model --dir ./my-model/Create Model
mixtrain model create <files...>Create a model from files. Supports directories and glob patterns.
Options:
| Option | Description |
|---|---|
--name, -n | Model name (defaults to first .py filename) |
--description, -d | Model description |
--entrypoint, -e | Entrypoint hint: file path, file:class, or class name |
# Single directory (most common)
mixtrain model create ./model-dir/ --name my-model
# Multiple files
mixtrain model create model.py utils.py --name my-model
# Glob patterns
mixtrain model create "*.py" Dockerfile --name my-model
# With description
mixtrain model create model.py -n my-model -d "Image classifier"
# With --entrypoint to select a specific class from a multi-class directory:
# File only - scan only this file, must contain exactly one MixModel class
mixtrain model create . --name diffusion --entrypoint models/diffusion.py
# File:Class - select a specific class inside a file
mixtrain model create . --name diffusion --entrypoint models/diffusion.py:DiffusionModel
# Class-only - class name must be unique across all uploaded files
mixtrain model create . --name classifier --entrypoint ClassifierEntrypoint hints
The --entrypoint (or -e) option lets you create a model from a specific class
when your source directory contains multiple MixModel classes. It supports three formats:
| Format | Example | Behaviour |
|---|---|---|
| File only | --entrypoint models.py | Scans only that file. Must contain exactly one MixModel class. |
| File:Class | --entrypoint models.py:DiffusionModel | Uses the named class in that file. |
| Class only | --entrypoint Classifier | Scans all files for a class with that name. Must be unique. |
Without --entrypoint, the historic behaviour applies: all .py files are scanned,
and exactly one MixModel class must exist across the entire upload.
Run Model
mixtrain model run <name> [--<input> <value> ...]Run a model. Streams logs in real-time for native models; external models (fal, openai, etc.) return immediately.
Each argument in the model's run function is can be passed via cli flags. Scalars, enums, and
resource references take a plain value; composite inputs (object, tuple, list,
dict, tensor) take a JSON value. Flag names accept dashes or underscores. Run
mixtrain model run <model-name> --help to see the available
arguments.
Options:
| Option | Description |
|---|---|
--<input> <value> | Set a spec input by name, e.g. --prompt "A sunset". Composite types take a JSON value. |
--input-json, -i | All inputs as one JSON object, or a path to a JSON file. Per-input flags override matching keys. |
--detach, -d | Start run and exit without streaming logs |
Inputs whose name collides with a built-in flag (e.g. image, timeout) can
only be set through --input-json; --help marks these.
# Per-argument flags
mixtrain model run my-model --prompt "A sunset" --steps 30
# All inputs as one JSON object or file
mixtrain model run my-model -i '{"prompt": "Hello"}' --detach
mixtrain model run my-model -i input.jsonView Model
mixtrain model get <name>View model details and recent runs.
mixtrain model get my-modelEdit Model
mixtrain model edit <name>View and modify model files and metadata.
Options:
| Option | Description |
|---|---|
--file, -f | Edit file in $EDITOR |
--view, -v | View file contents |
--delete, -d | Delete file/folder at this path |
--add, -a | Add local files to model |
--name, -n | Rename model |
--description, -D | Update description |
--yes, -y | Skip confirmation prompts |
mixtrain model edit my-model # List all files
mixtrain model edit my-model -v model.py # View file contents
mixtrain model edit my-model -f model.py # Edit file in $EDITOR
mixtrain model edit my-model --add ./src/ # Add files
mixtrain model edit my-model -d old.py # Delete file
mixtrain model edit my-model -n new-name # Rename model
mixtrain model edit my-model -D "Updated" # Update descriptionView Logs
mixtrain model logs <name> [run_number]View logs for a model run.
mixtrain model logs my-model # Latest run
mixtrain model logs my-model 5 # Run #5List Runs
mixtrain model runs <name>List all runs for a model.
mixtrain model runs my-modelDelete Model
mixtrain model delete <name>Delete a model.
Options:
| Option | Description |
|---|---|
--yes, -y | Skip confirmation |
mixtrain model delete my-model
mixtrain model delete my-model --yes