SUBSCRIPTIONS & PAY-AS-YOU-GO API

Start at $10.
Use every token well.

Subscribe when a person uses Cenmax AI. Choose PAYG when your application uses it. Every subscription includes Cenmax AI Desktop, Cenmax AI CLI, and an optional five-hour usage limit.

STARTER

$10/ month

For focused implementation work on a small, predictable monthly budget.

Choose Starter
  • Worker model access
  • 1× weekly usage allowance
  • Cenmax AI Desktop and Cenmax AI CLI
  • Optional five-hour usage limit
  • Up to 99.9% cache hits in long sessions
BEST FOR MOST DEVELOPERS

BUILDER

$50/ month

For developers who need Pro for difficult work and more room for long agent sessions.

Choose Builder
  • Pro and Worker model access
  • 5× the Starter weekly allowance
  • Cenmax AI Desktop and Cenmax AI CLI
  • Optional five-hour usage limit
  • Full, unnerfed model performance

POWER

$100/ month

For heavy daily use, larger codebases, and developers who keep several long tasks moving.

Choose Power
  • Pro and Worker model access
  • 10× the Starter weekly allowance
  • Cenmax AI Desktop and Cenmax AI CLI
  • Optional five-hour usage limit
  • Full, unnerfed model performance

Why Starter includes Worker only

Pro uses a larger allowance for difficult reasoning. On a $10 plan, only a few Pro sessions could consume the whole weekly quota. Worker can still plan and solve problems; Starter uses it so the plan stays useful across more sessions. Builder and Power unlock both models.

PAY-AS-YOU-GO API

Direct rates for software.

Use PAYG for applications, chatbots, internal tools, and automated agents. We charge cache-hit input at a lower rate, so an efficient long session costs less.

MODELINPUTCACHED INPUTOUTPUT
WorkerClear execution tasks
$1.50$0.30$5.00
ProPlanning, bugs, and difficult reasoning
$5.00$1.00$25.00

Current Cenmax AI rates · USD per one million tokens

Get API access Read the API quickstart

FLEXIBLE LIMITS

Use the five-hour limit—or turn it off.

The short window helps spread your allowance across the week. If you need one long session, disable it and use the weekly allowance at your own pace. Cenmax AI shows your usage and reset time either way.

LIMIT ON

Keep usage steady.

The five-hour window prevents one busy session from taking too much of the weekly allowance.

LIMIT OFF

Use the week your way.

Remove the short window when a long task matters more than spreading usage across several days.

COMMON QUESTIONS

Clear rules. No hidden policies.

Does the cache reduce my usage?

Yes. Cenmax AI Desktop and Cenmax AI CLI reuse stable prompt tokens. In long sessions they can reach a 99.9% cache hit rate, so repeated context uses much less quota than fresh input.

Is Worker a small or limited model?

No. Worker is not an ultra-small 30B-class model. It can understand a codebase, use tools, solve problems, and plan its own work. Pro simply handles deep planning and difficult reasoning better.

Why not put Pro in the $10 plan?

Pro uses more of the allowance during difficult reasoning. On a $10 plan, a few Pro sessions could consume the weekly quota. Worker still plans and solves problems; Starter uses it to keep the plan useful across more sessions.

Can I disable the five-hour limit?

Yes. The short window is optional. Turn it off when you want the weekly allowance to be your only usage limit.

Will Cenmax AI change limits without telling me?

No. We do not use hidden throttles or performance policies. The dashboard shows the limits and reset times that apply to your account.

Do developers need the API?

No. Subscriptions provide direct access through Cenmax AI Desktop and Cenmax AI CLI. The PAYG API is for software integrations.

Will the models slow down after I subscribe?

No. Cenmax AI does not intentionally nerf a model to push customers toward a new numbered release.