Netlify's credit-based pricing: what happens when the allowance runs out

Netlify has billed on a credit system since 4 September 2025. For anyone deploying static sites or Jamstack projects, that model matters more than the feature list, because it decides what happens when a project suddenly gets traffic. This is a summary of what the vendor documents and what users report. I have not run a hands-on test of Netlify myself.

How the credit system works

Instead of separate quotas for build minutes and bandwidth, Netlify bills five usage types against one shared monthly pool:

• Production deploys: 15 credits each
• Compute: 10 credits per GB-hour
• Bandwidth: 20 credits per GB
• Web requests: 2 credits per 10,000 requests
• AI inference: varies by model

According to Netlify (as of October 2026), deploy previews, branch deploys, failed deploys and rollbacks to an earlier production deploy use no credits. The review workflow is not penalized, which is a sensible design choice.

The real risk: a team-wide pause

Once a team's monthly allowance is used up, Netlify pauses every project on that team until the next billing cycle, not only the one that caused the spike. Notifications arrive at 50%, 75% and 100% of usage. Auto recharge can top up credits automatically, but it is off by default and only the Team Owner can enable it.

The Free plan is capped at 300 credits per month and never incurs charges. That protects you from a bill, not from an outage.

Practical tips

1. Do the bandwidth math before launch. At 20 credits per GB, 300 credits are 15 GB of bandwidth if nothing else is consumed. Personal includes 1,000 credits, which is 50 GB under the same assumption.
2. Decide on auto recharge deliberately. For a production site, either enable it or accept the pause risk knowingly.
3. Keep client sites and experiments on separate teams. The pause applies team-wide, so a hobby project should not share a credit pool with a client's site. This is my own inference from how the pause works.
4. Cap AI usage. Agent Runners run coding agents such as Claude Code, Codex CLI and Gemini CLI inside the dashboard. That usage is billed as AI inference, and an AI Credit Usage Limit stops new agent runs once it is reached.

Where Netlify fits, and where it does not

It fits Astro, Hugo, Eleventy, SvelteKit and other static or Jamstack setups, especially if you want a preview URL for every pull request. It is a weaker choice in three cases:

• Next.js with server-side rendering: new framework features usually reach Vercel first.
• Heavy full-stack apps: persistent backends and containers are better served by platforms such as Render or AWS.
• Unpredictable traffic: the credit model needs budget planning. Cloudflare Pages, with unlimited bandwidth on all plans, is the usual alternative here.

Accounts created before 4 September 2025 can stay on their legacy plan, according to Netlify, so check which model you are on before comparing numbers.

Takeaway

For frontend teams, the workflow is mature. The pricing model is the part that needs attention: know your bandwidth, set up notifications, and decide on auto recharge before the first traffic spike, not after it.

I run delavo.de, where this article was first published as part of a longer overview with a rating breakdown and FAQ: Netlify overview on delavo.de. The original draws on vendor documentation and user reports, not on a hands-on test. This summary was written with an AI agent based on that original.