Render Performance: Benchmarks, Latency & Limits 2026
Render is instance-priced from a 512 MB Free tier to a 32 GB Pro Ultra at $450/mo, scaling to 100. Free spins down after 15 minutes with a ~1-minute cold start.
Render Performance verdict
Render is instance-priced, not concurrency-capped.
Web services run from a 512 MB Free tier up to a 32 GB Pro Ultra at $450 a month, scaling horizontally to 100 instances. Autoscaling on a target CPU or memory percentage is gated to the Pro plan.
Never run production on Free, where the 15-minute spin-down and roughly 1-minute cold start are disqualifying. Start at Starter or Standard for always-on. Size the instance tier to your memory and CPU, and budget for autoscaling on Pro, which is instances times the tier price, prorated by the second. Keep persistent disks off any service you need to scale, because a disk pins it to a single instance.
- OpenStatus tested the free tier, where cold starts after the 15-minute spin-down pushed per-region p99 toward 16s. Paid instances stay always-on and avoid this.
- Autoscaling on a CPU or memory target requires the Pro plan or higher, so lower plans scale manually only. A service with an attached persistent disk cannot scale beyond one instance.
- Contractual uptime SLAs are Enterprise-only, so lower plans have no guaranteed uptime.
- Avg latency (independent)
- 451 ms
- Max instances
- 100
- Free spin-down
- 15 min
- Top instance
- 32 GB / 8 CPU
- Contractual SLA
- Enterprise
This page covers how fast Render runs, how it scales and what it costs. Region coverage lives on its own page.
Size your Render service and autoscaling
- At 85% observed CPU against a 60% target, autoscaling runs 3 × Starter (ceil of 2 × 85/60).
- That is about $21/mo at 3 × $7 (Render prorates compute by the second).
Estimated from Render's published instance prices and autoscaling rule (ceil(current x observed / target)). Autoscaling needs the Pro plan or higher; below it you set instances manually. Verify against your own workload.
Render measured latency, independent data
| Metric | Value | Sample |
|---|---|---|
| Average latency | 451 ms | free, Frankfurt, 10,946 pings |
| p75 latency | 447 ms | OpenStatus, Feb 2024 |
| p90 latency | 591 ms | OpenStatus, Feb 2024 |
| p95 latency | 707 ms | OpenStatus, Feb 2024 |
| Warm TTFB, best probe | 107 ms | Amsterdam to Frankfurt |
| Warm TTFB, worst probe | 720 ms | Johannesburg to Frankfurt |
| Uptime, two weeks | 99.89% | free tier, 12 fails / 10,946 pings |
Render scaling and autoscaling
| Capability | Value | Notes |
|---|---|---|
| Max instances per service | 100 | Horizontal scale ceiling |
| Manual scaling | All plans | Set a fixed instance count |
| Autoscaling | Pro plan or higher | Scales on target CPU and/or memory % |
| Autoscaling formula | ceil(current x current/target) | new_instances from utilization ratio |
| Scale direction | Up now, down later | Scales up immediately, waits minutes to scale down |
| Persistent-disk services | Single instance | Cannot scale to multiple instances |
| Scaling billing | Prorated by the second | No extra cost for a scaling action |
Instance tiers
| Plan | RAM / CPU | Price |
|---|---|---|
| Free | 512 MB / 0.1 CPU | $0/mo |
| Starter | 512 MB / 0.5 CPU | $7/mo |
| Standard | 2 GB / 1 CPU | $25/mo |
| Pro | 4 GB / 2 CPU | $85/mo |
| Pro Plus | 8 GB / 4 CPU | $175/mo |
| Pro Max | 16 GB / 4 CPU | $225/mo |
| Pro Ultra | 32 GB / 8 CPU | $450/mo |
| Custom / Enterprise | Up to 512 GB / 64 CPU | Contact sales |
Render reliability and architecture
- Contractual uptime SLAs are only available on the Enterprise plan; lower plans carry no published uptime guarantee
- Render's infrastructure runs on Amazon Web Services and Google Cloud Platform, with Cloudflare providing the CDN and DDoS layer
- Every application on Render gets free, automatic DDoS protection through Cloudflare with no configuration required
- Free, fully-managed TLS certificates are included for all web services
- Render is certified for SOC 2 Type 2 and ISO 27001, offers a GDPR DPA, and supports HIPAA on the Scale plan and above
- Free web services spin down after 15 minutes without inbound traffic and take about one minute to spin back up on the next request
Render latency benchmarks, independently measured
- OpenStatus independently monitored a basic Hono server on Render's free tier in Frankfurt every 10 minutes from six global locations, gathering 10,946 pings over two weeks in February 2024
- Render averaged 451ms with 99.89% uptime in the run, behind Cloudflare Workers (182ms) and Railway (381ms) but ahead of Koyeb (539ms) and Fly.io (1,471ms cold)
- Per-region p99 latency spiked from 15,979ms to 16,696ms, which OpenStatus attributed to occasional free-tier cold starts taking up to about 50 seconds against a 30s timeout
- Warm time-to-first-byte ranged from 107ms (Amsterdam) to 720ms (Johannesburg), tracking distance from the probe to the single Frankfurt region
- Edge caching for paid web services is powered by the same global Cloudflare CDN as Render static sites, serving cached assets close to users
- Free instances cannot use edge caching, so the benchmark's free-tier latency reflects origin responses, not CDN-cached delivery
Render Performance FAQ
How fast is Render, and what do independent benchmarks show?
OpenStatus independently monitored a free Render service in Frankfurt every 10 minutes from 6 global probes over two weeks in Feb 2024. It averaged 451ms, 447ms p75, 707ms p95, and 99.89% uptime. Warm TTFB ran 107ms from Amsterdam up to 720ms from Johannesburg. Per-region p99 spiked near 16s, because free-tier cold starts can take up to 50 seconds, which paid always-on instances avoid.
How does scaling and autoscaling work on Render?
Render scales horizontally to a maximum of 100 instances per service. Manual scaling at a fixed instance count is on every plan, while autoscaling on a target CPU or memory percentage requires the Pro plan or higher. The formula is new_instances = ceil(current_instances × current_util / target_util). Render scales up immediately and waits a few minutes before scaling down. A service with a persistent disk cannot scale past one instance.
What instance tiers and prices does Render offer?
Tiers run from Free at 512 MB and 0.1 CPU for $0, Starter at 512 MB and 0.5 CPU for $7, Standard at 2 GB and 1 CPU for $25. Pro is 4 GB and 2 CPU at $85, Pro Plus 8 GB and 4 CPU at $175, Pro Max 16 GB and 4 CPU at $225. Pro Ultra tops out at 32 GB and 8 CPU for $450 a month. Custom Enterprise instances reach 512 GB RAM and 64 CPU. Compute is prorated by the second.
Do Render services have cold starts?
Only the Free tier. Free web services spin down after 15 minutes without inbound traffic and take about one minute to spin back up on the next request, showing a loading page meanwhile. Paid instances stay always-on, so they have no spin-down cold start. That is why the independent benchmark's free-tier p99 spiked to about 16s.
What reliability and compliance does Render provide?
Render runs on AWS and Google Cloud with free, automatic Cloudflare DDoS protection and managed TLS on every web service. It is certified for SOC 2 Type 2 and ISO 27001, offers a GDPR DPA, and supports HIPAA on the Scale plan for a 20% compute premium. Contractual uptime SLAs, though, are only available on the Enterprise plan.
Sources & verification
| Source | What was checked | Last checked |
|---|---|---|
| Render Official | Official product page | July 10, 2026 |
| OpenStatus | Independent latency benchmark | July 10, 2026 |
| Render Ddos Protection | Ddos Protection | July 10, 2026 |
| Render Free | Free | July 10, 2026 |
| Render Pricing | Pricing and plans | July 10, 2026 |
| Render Scaling | Scaling and concurrency | July 10, 2026 |
| Render Security | Security and compliance | July 10, 2026 |
Every fact on this Render page is tied to a named source and a verification date. Freshness-sensitive figures trace to the sources above; verify against the vendor before relying on them.
Explore Render
Every page on Render in one place, you are on performance.