Kimi Code

z.ai coding plan limits

How z.ai coding plan limits work: the quota window, model access, and the difference between plan quota and API rate limits.

Quick answer

The z.ai coding plan grants a quota over a short window rather than a monthly message count. Consumption depends on the model and context, so a long agent session drains it faster than short chats. API keys are metered separately with their own rate limits.

Structure

ElementApplies toNotes
Coding plan quotaPlan usage in the editorShort window
Model accessIncluded modelsVaries by tier
API rate limitsKeys and scriptsPer minute
BillingSubscriptionPer period

What drains the quota

  • Long agent runs with full repository context.
  • Repeated retries after an error.
  • Premium model selection for simple edits.
  • Parallel sessions competing for the same window.

Decide whether the plan fits

  1. Measure one representative task.
  2. Estimate how many you complete per day.
  3. Compare with the quota the plan provides.
  4. If you hit the window daily, the tier is too small.
  5. If you never approach it, you are paying for headroom.

For the cost comparison with a first-party plan, see z.ai vs Claude Code cost.

Plan limits in practice

The coding plan is sized for an editor session, not for parallel agents. Two windows competing for the same quota is the most common cause of an early block, and it looks like a billing fault because each session seems modest on its own.

HabitEffect on quotaChange
One agent at a timePredictableKeep
Parallel agentsDrains fastSerialize
Full-repo contextHigh input per turnScope the files
Repeated retriesMultiplies callsStop and replan

See z.ai vs Claude Code for the cost comparison and AI rate limits for the window shapes.

Questions, answered

Is the z.ai coding plan a monthly message count?

No. It is a quota over a short window, and consumption depends on model and context.

Do API keys share the coding plan quota?

No. API usage is metered separately with its own rate limits and billing.

Does the plan include every model?

Model access varies by tier. Check the plan page for the current list.

Browse by topic