How billing works
How do I pay for FlexAI?
Pay as you go. You’re billed for what you actually use — per token for inference, per GPU-hour for training and fine-tuning — at the rates published on flex.ai/pricing. There’s no subscription, and no commitment beyond your usage.Do I need a card to get started?
A card is required to create an API key, but not to sign up. You can create an account, browse the model catalog, and run the in-browser demos on flex.ai without adding one. When you’re ready for a key, add your billing address and card from the billing page.Where credits come from
How do I get FlexAI credits?
Credits arrive with a program or an agreement rather than automatically at signup:- FlexAI Startup Program — accepted startups receive $1,000 in Token Factory credit, then 50% off their first three months and 30% off the next three. Apply on flex.ai/startups.
- Negotiated and invited offers — for an evaluation, a proof of concept, or a committed-use agreement, talk to commercial@flex.ai.
New accounts do not come with an automatic credit grant. If you are looking for the monthly free-credit offer advertised previously, it was retired on 2026-08-20.
Are existing users eligible for credits?
Yes — credits aren’t limited to new accounts. Reach out to commercial@flex.ai with your workload and we’ll tell you what applies.Using credits
How is credit usage calculated?
Credits draw down against your usage at the same published rates you’d otherwise pay: per million input and output tokens for inference, per GPU-hour for training and fine-tuning. Per-model rates are in the model catalog, and the full picture is on flex.ai/pricing. Your balance draws down before your card is charged, so a credited account bills nothing until the credit is used.Do credits expire?
Each grant carries its own terms, including any expiry, and those terms are stated when the credit is applied to your account. Check the billing page for your current balance, or ask commercial@flex.ai if you’re unsure which terms apply to you.What happens when my credits run out?
For the Inference API, requests return402 Payment Required and start succeeding again as soon as you top up — no key regeneration needed. See Billing & Quotas for the exact response and how to add funds.
For training and fine-tuning workloads, talk to us before you run out and we’ll help you plan the transition.