Models

A model in PiG is identified by a provider-qualified spec: provider/modelID. Every API that takes a model - --model, /model, Ctrl+P cycling, Context.SetModel(), settings entries - should round-trip through provider-qualified specs. Bare model IDs are accepted for backward compatibility but are interpreted as openai/<id>, which has historically caused silent routing drift.

Selecting a model

The selector lists models from providers with configured authentication. It uses GEMINI_API_KEY for Google Gemini, not GOOGLE_API_KEY. Providers without configured authentication are omitted.

Scoped models

Scoped models are the list Ctrl+P cycles through. They are stored in settings as provider-qualified IDs; PiG accepts legacy bare IDs and rewrites them on next save.

Cycling preserves both provider and id. The generatedModelSpec helper produces specs like:

• github-copilot/gpt-5.5
• openai/gpt-5.5-mini
• openrouter/openai/gpt-5.5
  • trailing slash inside id is preserved.

If a custom helper or extension passes only model.ID to BuildModel(), PiG falls back to openai/<id> - never do this in cycling code paths.

Thinking levels

PiG surfaces reasoning effort through these named levels:

off → minimal → low → medium → high → xhigh → max

A model exposes only the levels it supports. Cycling follows that supported set.

• Shift+Tab cycles thinking on models that advertise reasoning support.
• getThinkingLevel() and setThinkingLevel(level) are exposed to extensions.
• The current label is rendered in the status line; extensions can override the hidden label with ui.setHiddenThinkingLabel.
• Models that do not support reasoning silently ignore changes; the host does not emit an error.

A model is reasoning-capable when its generated entry sets Reasoning: true; ThinkingLevelMap controls how each level is wired into the request. See ai/models_generated.go in source for the table.

Model metadata

getAllTools() excluded, every model entry the host knows about carries:

• Provider
  • provider key (see providers.md).
• ID
  • model identifier as the provider names it. May contain slashes.
• DisplayName
  • UI label.
• Reasoning
  • bool; whether the model supports thinking_level.
• ThinkingLevelMap
  • provider-specific wiring per level.
• MaxTokens / ContextWindow
  • for context usage math.
• Cost (if known) - input/output rates.

Extensions can read this via getModelInfo() (returns *ModelInfo) inside any handler.

Adding a model

New built-in models are generated into ai/models_generated.go and committed in source. Extensions cannot add new built-in models, but they can register an independent provider and model catalog. See Providers and Extensions.

Common errors