- Dufus powers the AI coding agent. Its models are named after Earth’s oceans and are optimized for reasoning over code, repos, and tool use.
- Igno powers the assistant and automation agent. Its models are named after cloud types and are optimized for conversation, planning, and everyday task execution.
Dufus: coding models
Use Dufus models when you want code written, refactored, reviewed, or debugged, or when you need an agent to plan and execute multi-step work across a repository.Arctic
The fastest and cheapest Dufus model. Good for autocompletion-style asks, quick “what does this function do” questions, and small localized edits.- Latency: very low
- Tool use: basic (read files, run short commands)
- Recommended when: you want speed and are working in a single file
Antarctic
The default Dufus model for most users. Handles typical day-to-day coding work well without the flagship cost.- Latency: low
- Tool use: full plugin support
- Recommended when: you are reviewing PRs, writing features, or refactoring within a module
Atlantic
Stronger reasoning and a larger context window. Better at holding a multi-file change in its head and following longer plans.- Latency: medium
- Tool use: full plugin support with parallel calls
- Recommended when: a task spans several files, needs careful debugging, or requires a plan before edits
Pacific
The flagship Dufus model. Highest capability, largest context, and the deepest reasoning budget. Use it when other models get stuck.- Latency: higher
- Tool use: full plugin support with parallel calls and long agent loops
- Recommended when: you are doing whole-repo migrations, hard architectural work, or long autonomous sessions
Igno: assistant and automation models
Use Igno models for chat, research, planning, scheduling, reminders, and running automations. Igno models are tuned for natural conversation, structured outputs, and reliable tool use across connected integrations.Cirrus
The lightest Igno model. Fast, cheap, and good enough for most short conversational turns.- Latency: very low
- Tool use: basic
- Recommended when: you want a quick answer or a simple reminder
Stratus
The default Igno model. A solid balance of quality and cost for daily assistant use.- Latency: low
- Tool use: full plugin support
- Recommended when: you are summarizing content, drafting messages, or running standard automations
Nimbus
Stronger reasoning and a much larger context window. Good for research tasks that need to hold many sources at once.- Latency: medium
- Tool use: full plugin support with parallel calls
- Recommended when: you are doing deep research, comparing long documents, or orchestrating several plugins in one automation
Aurora
The flagship Igno model. Highest capability across reasoning, planning, and long-context understanding.- Latency: higher
- Tool use: full plugin support with parallel calls and long agent loops
- Recommended when: the task is high-stakes, ambiguous, or requires long autonomous execution
Choosing between Dufus and Igno
You can switch models any time from the model picker at the top of a conversation. Automations remember the model they were created with, so change the model in the automation settings if you want it to use a different tier.
Pricing notes
- All prices above are per 1M tokens and are billed separately for input and output.
- Tool use, plugin calls, and file uploads may incur additional per-call costs depending on the plugin.
- Enterprise plans can pin defaults per workspace and set spend limits per model.