Model Catalog & Reasoning

Specifying Models

You can select a model when starting term2 using the -m or --model flag:

# Model identifier
term2 -m gpt-5.1

# Provider and model combination
term2 -p openrouter -m anthropic/claude-3.7-sonnet

# Model with reasoning effort specified inline
term2 -m o3-mini:high

In an interactive session:

• Run /model <model-name> to switch models immediately.
• Press Ctrl+O or run bare /model to open the interactive Model Picker menu.

Listing Available Models (--list-models)

term2 can query the live catalog of models available across all configured providers:

# List all available models
term2 --list-models

# Filter by a search term
term2 --list-models sonnet
term2 --list-models gpt-5

# Bypass disk cache and fetch fresh models from provider APIs
term2 --list-models --refresh

Reasoning Effort (-r, --reasoning, /effort)

For models that support variable reasoning effort (such as OpenAI o1, o3, Grok, and thinking models), configure the reasoning level:

Supported levels: default, none, minimal, low, medium, high, xhigh.

# Set at launch
term2 -r high

# Set in an interactive session
/effort medium

# Or open the reasoning menu directly with keyboard shortcut
Ctrl+T

default indicates that term2 will not send an explicit reasoning parameter to the API, allowing the provider or model default to take effect.

Favorites & Nicknames

In the interactive model picker (Ctrl+O):

• Favorites (Ctrl+F): Mark frequently used models as favorites. Favorited models appear at the top of the picker and resolve instantly.
• Nicknames (Ctrl+N): Assign a short name to a model (e.g. sonnet -> openrouter/anthropic/claude-3.7-sonnet). Passing the nickname to -m resolves immediately without waiting for catalog lookups.