GeezerVison is showing ampcode.com — Modes & Models

Modes & Models

The Amp coding agent uses the best model for each task: leading generalist foundation models for complex reasoning and planning, and smaller specialized models for fast, accurate responses in specific domains.

Agent Modes

Low

Fast, low-cost mode for small, well-defined tasks

Agent GLM-5.3-Flash · high

Medium

Default for most tasks, balancing quality, speed, and cost

Agent Opus 5.5 · high

High

Use for difficult tasks that need deeper reasoning and can tolerate more time and cost

Agent GPT-6 Astra · med

Ultra

Use for the hardest open-ended tasks when maximum capability matters more than speed or cost

Agent Fable 5.2 · high

Puck

Navigate work across Amp projects, threads, and context

Opus 5.5 or GPT-6.1 Sol

Puck Auto uses Opus 5.5 with an active, unpaused Claude subscription when personal Model Routing is allowed. Otherwise it uses GPT-6.1 Sol. Both default to low effort. A usable saved model pin overrides Auto.

Subagents

Task

Delegated coding, research, and verification work

Same as Thread

Read Thread Reading & summarizing other Amp threads Same as Thread

System Models

Dictation Voice dictation of prompts and messages GPT Transcribe

Call Puck

Voice calls with Puck

GPT Live 1

View Media Analysis of images, PDFs, audio, and video Gemini 3.7 Flash

Painter

Image generation and editing

GPT Image 2.5 Sunburst

Titling Title generation for threads Same as ThreadCompaction Context summarization for long threads Same as Thread

See the Amp Security Reference for more details on model use and inference providers.