⌨️ CLI Usage

Pass any flag to skip the TUI and get a non-interactive result. Combine flags to filter, sort, and shape the output for pipelines.

models-dev --search claude
models-dev --provider openai --sort input-cost
models-dev --tool --reasoning
models-dev --provider anthropic --compact
models-dev --search grok --json

🚩 Flags

• --search <term>: substring match on name/id
• --provider <name>: provider id or display name (case-insensitive)
• --tool: only models with tool calling
• --reasoning: only models with reasoning capability
• --sort <field>: one of input-cost, output-cost, provider (the TUI exposes additional sorts)
• --json: emit raw JSON for the resulting models
• --compact: print provider:model:name:id per line
• --ui <mode>: blessed | table | auto (default)

📊 Output Columns

Shown in both the non-interactive table and the interactive list view when applicable:

• Provider, Model, Provider ID, Model ID
• Tool, Reason (reasoning capability)
• Modalities (input/output)
• Costs (input/output/cache read/cache write; per 1M tokens)
• Limits (context, output tokens)
• Temperature flag, Weight openness, Knowledge cutoff, Release, Updated

Missing numeric/cost fields render as -.