Docs /Getting Started/Developer and Credits

Developer and Credits

How prepaid credits work for Developer API billing.

The Developer plan is Featherless's usage-based plan: you choose a monthly credit amount (from $50/month) and API usage is billed per token against your balance. Context sizes up to 256K, the fastest response times, one included agent environment, and no model-size limit. Unused credits roll over and don't expire.

How Developer Pricing Works

  • Choose a monthly credit amount.

  • That amount is added to your Featherless credit balance each billing cycle after payment succeeds.

  • You can add a one-time credit top-up from the billing page without changing your monthly credit amount.

  • Successful API requests reduce your balance.

  • If your balance reaches zero or below, request-pricing API calls are blocked until more credits are available.

  • Credits do not expire.

How Request Costs Are Calculated

  • Only successful requests are charged.

  • Request cost is based on the model used, input tokens, and output tokens.

  • Formula: input tokens x input price + output tokens x output price.

  • Prices are listed per 1M tokens.

How cached input pricing works

Some models have a lower "cached input" rate alongside their standard input rate. When a request reuses a long prompt prefix the platform has recently served — a long system prompt, running chat history, a shared document — eligible input tokens may be billed at the cached rate instead of the full input rate.

  • It's automatic: no code changes, no cache parameters.

  • Only sufficiently long repeated prefixes qualify (roughly 1,000+ tokens of exact prefix match). Short or novel prompts bill at the standard input rate.

  • Each model's page shows its cached rate when it has one (example: GLM-5.3 — $1.40 input / $0.26 cached / $4.40 output per 1M tokens). A "–" means the model always bills at the standard input rate.

  • For agentic and long-context workloads, cached input pricing can cut input costs substantially.

What each request costs

Every model's page shows its current prices — input, cached input where available, and output, per 1M tokens (per 1M characters for speech models): https://featherless.ai/models

Cost per request = input tokens × input price + output tokens × output price. Only successful requests are charged. Some low-volume niche models carry a pricing multiplier over their model class's base rate the model page always shows the effective price if you are on Developer.

Live Balance And Recent Usage

  • Your billing page shows your up-to-date live credit balance.

  • You can also read the same live balance and usage activity as the billing page with an organization Admin API key. See the /credits API reference and the /usage/activity API reference.

  • Recent usage may take a short time to appear in detailed usage views.

  • Usage activity shows spend, request count, input and output tokens, models used, and request-level details.

  • The billing page also lets eligible request-pricing organizations add one-time credits using the saved billing method on file.

Low Balance And Out-Of-Credits Emails

  • We send a low-balance email when your credit balance reaches $5.

  • We send an out-of-credits email when your balance reaches zero or below.

  • Alerts are sent to the organization billing email.

  • Alert settings can be changed from your billing page.

Choosing A Monthly Credit Amount

  • Start with the amount you expect to spend in a month.

  • You can increase your monthly amount later if usage grows.

  • If you are unsure, start with a smaller tier and monitor usage from the billing page.

  • For high-volume usage, you can add one-time credits from the billing page or contact support for custom needs.

Changing Your Monthly Credit Amount

  • Increasing your monthly credit amount takes effect immediately after successful payment.

  • The upgrade is prorated for the remaining time in your current billing cycle.

  • Your billing cycle date stays the same.

  • Decreasing your monthly credit amount takes effect immediately and is not prorated; the lower amount is billed at your next billing cycle.

Switching From A Chat Plan To Developer

  • Switching from a chat monthly plan to Developer changes your organization to credit-based billing.

  • The change can take effect immediately after successful payment.

  • You can switch to any monthly credit amount, whether higher or lower than your current plan's monthly price.

  • Switching does not prorate: the unused portion of your current fixed-plan period is not credited or refunded, and you start a fresh credit billing cycle.

  • Your billing cycle date stays the same where possible.

Switching From Developer To A Chat Plan

  • You can switch back to a fixed monthly chat plan.

  • Unused credits stay on your organization and do not expire.

  • Stored credits are only used while your organization is on request pricing.

Cancelling Developer

  • Cancelling takes effect at the end of your current billing period: monthly credit top-ups stop and your organization then moves to the free Developer base.

  • Existing credits remain available until used.

  • Once your credit balance reaches zero or below, request-pricing API calls are blocked unless you restart request pricing or add more credits.

Refunds And Adjustments

  • Refunds or account adjustments may reduce your credit balance.

  • Contact support if you have questions about a balance change.

Last edited: Sep 15, 2026