Skip to content

Models Commands

Browse, search, and download models from the models.dev catalog and Hugging Face Hub.

Results are cached at ~/.superqode/models_cache.json with a 1-hour TTL. Use --refresh to force an update. Use --live to query the provider's /v1/models endpoint directly instead of the catalog.

models (default)

Browse or search the models.dev catalog of 5000+ models from 130+ providers.

superqode models [OPTIONS]

Options

Option Description
--search, -s Search query (name, provider, or description)
--provider, -p Filter by provider ID
--cap Filter by capability: tools, vision, reasoning, code, long, json
--free Show only free (no-cost) models
--max-price Maximum price per million input tokens
--curated Show only curated/recommended models
--sort Sort order: provider, price, context
--limit Maximum results (default 50, 0 for all)
--refresh Bypass cache and fetch fresh data
--live Query provider's /v1/models endpoint directly
--json Emit JSON output

Examples

# Browse default view
superqode models

# Search for models
superqode models -s claude

# Filter by provider and capability
superqode models -p anthropic --cap tools

# Free models with reasoning
superqode models --free --cap reasoning

# Sort by context window
superqode models --sort context --limit 10

# Live query of provider endpoint
superqode models -p openai --live --json

# All models, no limit
superqode models --limit 0

models providers

List all providers available in the models.dev catalog.

superqode models providers [OPTIONS]

Options

Option Description
--json Emit JSON output

Examples

superqode models providers
superqode models providers --json

models show

Show detailed information for a specific provider/model combination: context window, pricing, capabilities, and required API key environment variables.

superqode models show PROVIDER/MODEL

Arguments

Argument Description
PROVIDER/MODEL Provider and model ID (e.g., anthropic/claude-sonnet-4)

Examples

superqode models show anthropic/claude-sonnet-4
superqode models show openai/gpt-4o

models hub

Search the Hugging Face Hub for models.

superqode models hub [QUERY] [OPTIONS]

Arguments

Argument Description
QUERY Optional search query

Options

Option Description
--gguf Filter for GGUF format models
--mlx Filter for MLX format models
--sort Sort order: downloads, likes, trending_score, created_at
--limit Maximum results (default 25)
--json Emit JSON output

Examples

# Search Hugging Face
superqode models hub qwen

# Filter by format
superqode models hub --gguf qwen
superqode models hub --mlx deepseek

# Sort by downloads
superqode models hub --sort downloads --limit 10

# Full JSON output
superqode models hub --json

models download

Download a model from Hugging Face Hub.

superqode models download REPO_ID [OPTIONS]

Arguments

Argument Description
REPO_ID Hugging Face repository ID (e.g., QuantFactory/Qwen3-8B-GGUF)

Options

Option Description
--to Target format: auto, ollama, mlx, transformers
--quant Quantization (default Q4_K_M)
--dir Target download directory
--name Custom model name for registration
--register / --no-register Register model with superqode after download
--yes Skip confirmation prompts

Examples

# Download with defaults
superqode models download QuantFactory/Qwen3-8B-GGUF

# Download for Ollama with custom name
superqode models download QuantFactory/Qwen3-8B-GGUF --to ollama --name qwen3-8b

# Download to specific directory, no register
superqode models download mlx-community/Qwen2.5-Coder-3B-4bit --to mlx --dir ~/models --no-register

# Non-interactive
superqode models download QuantFactory/Qwen3-8B-GGUF --yes

models convert-mlx

Convert a Hugging Face model to MLX format.

superqode models convert-mlx HF_PATH [OPTIONS]

Arguments

Argument Description
HF_PATH Hugging Face model path

Options

Option Description
--q-bits Quantization bits: 4 or 8
--no-quantize Skip quantization (keep float16/float32)
--out Output directory
--upload Upload result to Hugging Face Hub

Examples

# Convert with 4-bit quantization
superqode models convert-mlx Qwen/Qwen3-8B --q-bits 4

# Convert without quantization
superqode models convert-mlx Qwen/Qwen3-8B --no-quantize --out ./mlx-model

# Convert and upload
superqode models convert-mlx Qwen/Qwen3-8B --q-bits 8 --upload

models cached

List models cached locally from Hugging Face.

superqode models cached [OPTIONS]

Options

Option Description
--json Emit JSON output

Examples

superqode models cached
superqode models cached --json

models rm

Delete cached models matching a pattern.

superqode models rm PATTERN [OPTIONS]

Arguments

Argument Description
PATTERN Pattern to match cached model names

Options

Option Description
--yes Skip confirmation prompts

Examples

# Remove matching cached models (with confirmation)
superqode models rm qwen

# Force remove without confirmation
superqode models rm qwen --yes