Pick a ceiling,
then forget it.
Every subscription includes Kaeonix Chat and Kaeonix Code on the full Avatar model. Plans differ in how hard you can push a session and how much you can run in a month — pick a ceiling once and the monthly spend stays flat.
Monthly volume as a share of Max, the largest plan. The session window — how hard a single run can push — scales alongside it.
Pro
Full Avatar model, agent tools, whole-repo context — with enough headroom for a working day of real tasks.
- Chat and Kaeonix Code included
- 250K tokens per session window
- 4.5M tokens a month
- Sandboxing, plan mode, subagents
Ultra
Everything in Pro, with room for longer sessions and bigger refactors — more of them, before limits enter the picture.
- Everything in Pro
- 600K tokens per session window
- 12M tokens a month
- Priority capacity at peak hours
Max
Everything in Ultra, with the highest session ceiling and volume — built for leaving the agent running across large codebases all day.
- Everything in Ultra
- 1M tokens per session window
- 20M tokens a month
- Headless and CI runs at scale
What separates the plans.
| Feature | PRO$20 | ULTRA$50 | MAX$100 |
|---|---|---|---|
| Avatar model | Full | Full | Full |
| Kaeonix Chat | Included | Included | Included |
| Kaeonix Code | Included | Included | Included |
| Session window | 250K tokens | 600K tokens | 1M tokens |
| Monthly volume | 4.5M tokens | 12M tokens | 20M tokens |
| Peak-hour capacity | Standard | Priority | Priority |
| Headless / CI runs | Light | Yes | At scale |
| Sandboxing & plan mode | Yes | Yes | Yes |
| API access | Usage-based | Usage-based | Usage-based |
Usage-based access for your own agents.
A FeynCode-issued key gives Avatar to anything you build — your editor, your pipeline, or Kaeonix Code pointed at API billing instead of a plan. Billed per million tokens, with cached input charged at a reduced rate.
Rates are for Avatar 1.0; at the input rate a dollar buys 1M tokens. Limits and streaming details live in the docs.
API DOCS- What counts against my plan?
- Model work — the tokens Avatar reads and writes across chat and terminal sessions. Cached context is charged at a lower weight, which is why long sessions stretch further here.
- Can I use one plan for chat and the CLI?
- Yes. One FeynCode account and one balance covers both surfaces, on any tier.
- Do you support bringing my own keys?
- Kaeonix runs on Avatar, our own model, so access comes from a plan or an API key we issue. That is what lets us tune the agent and the model together.
- What happens when I hit a ceiling?
- Sessions slow rather than stop, and you can move up a tier or switch to API billing mid-month with the difference prorated.
- Is there a team plan?
- Max covers small teams sharing agent workloads. For seats, invoicing, and pooled volume, talk to us.
Start on Pro.
Move when it pinches.
Change tier or switch to API billing at any time; the account and the history stay the same.