> ## Documentation Index
> Fetch the complete documentation index at: https://upstash.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Scale to Zero

> Pay nothing when idle, without giving up capacity when traffic is high.

Traditionally, cloud services required users to predict their resource needs and provision servers based on those predictions. This often led to over-provisioning for potential peak loads and paying for unused resources during periods of low demand.

Upstash solves this with two pricing models:

* **Pay-as-you-go** databases *scale to zero*: you're billed per command, and an idle database costs \$0.
* **Fixed plans** offer a flat monthly price with unlimited commands for sustained, high-throughput workloads.

Both run on the same high-performance production infrastructure.

## Pay for usage

On pay-as-you-go, you're only charged for the commands you run (\$0.20 per 100K commands). When your application has little or no traffic, you pay little or nothing.

This is ideal for:

* Small teams and side projects that don't want a fixed monthly bill
* Spiky traffic, where provisioning for peak would mean paying for idle capacity most of the time
* Prototypes and internal tools with low or intermittent usage

## Fixed plans for high throughput

Once your traffic is sustained, per-command billing may no longer be the best fit. Fixed plans start at \$10/month and offer:

* **Unlimited commands** at a flat monthly price, so costs stay predictable no matter how much traffic you serve
* **Higher throughput limits**: up to 16,000 commands per second on larger fixed plans, and 100K+ commands per second with dedicated resources on Enterprise
* **Graceful behavior under load**: requests beyond your plan's throughput limit are briefly queued instead of rejected, so traffic spikes don't turn into errors

See [Redis pricing](https://upstash.com/pricing/redis) for the full plan comparison, or a [live throughput benchmark](https://upstash.com/blog/upstash-vs-aws-elasticache-serverless-redis-pricing-and-performance-2026) sustaining 16,000+ commands per second on a fixed plan.

## Scale up, not just down

Because scaling is automatic in both directions, you never manage capacity yourself. During traffic spikes, Upstash scales up to meet demand; during quiet periods, resources (and your bill, on pay-as-you-go) scale back down.

You can start on the free tier, move to pay-as-you-go as you grow, and switch to a fixed plan once sustained throughput makes a flat price the cheaper, more predictable option.


## Related topics

- [Compare](/docs/redis/overall/compare.md)
- [Pricing & Billing](/docs/redis/overall/billing.md)
- [Express Session with Serverless Redis](/docs/redis/tutorials/express_session.md)
- [Build Stateful Applications with AWS App Runner and Serverless Redis](/docs/redis/tutorials/aws_app_runner_with_redis.md)