The trial credit numbers look generous — $200, $300, even free forever — until you find out what runs out first and what the bill looks like the morning after.
Four providers. Four very different answers to "how long can I build before I pay?" Vultr hands you $300 in credits with no expiration date. DigitalOcean gives you $200 good for 60 days. AWS restructured its entire free tier model in July 2025 and now offers up to $200 for new accounts, with the clock running out at six months regardless of balance. Railway skips the trial credit model entirely: $5 flat per month, with $5 of usage credit included, so a small service costs you nothing beyond the plan fee.
None of these are marketing gimmicks. They are genuinely useful runway. But they work differently, they expire differently, and what happens after they expire is where most developers get surprised.
AWS accounts created after July 15, 2025 no longer get the classic 12-month free tier. Instead, there are two options at signup: Free Plan and Paid Plan. Free Plan accounts receive up to $200 in AWS credits and are shut down at six months regardless of remaining credit balance. Paid Plan accounts also receive up to $200 in credits, are billed normally once credits are exhausted, and remain open indefinitely.
There is no permanent free compute tier for new accounts. The AWS Always Free tier still exists for certain services (Lambda, DynamoDB, CloudWatch at limited scales) but there is no free EC2, no free RDS, and no free Lightsail past the trial window.
The $200 credit goes fast if you run anything with a real workload. A t3.small EC2 instance plus an RDS db.t3.micro Multi-AZ plus 100 GB of EBS storage runs roughly $80 to $100 per month at list price, so you are looking at about two months of real infrastructure. Legacy accounts created before July 15, 2025 are unaffected and retain 12-month free tier allowances. For a complete breakdown of both models, see the AWS Free Tier in 2026 guide.
DigitalOcean gives new accounts $200 in credits valid for 60 days. That is a hard expiration: unused credits disappear at day 61. The $200 is enough to run a meaningful stack for two months, including a Droplet, a managed database, and object storage with room to spare.
After the trial, DigitalOcean's pricing is clean and predictable. The smallest Droplet is $6 per month: 1 vCPU, 1 GB RAM, 25 GB SSD, and 1 TB of outbound bandwidth included. That 1 TB is the important part. There is no egress billing surprise. You run your app, traffic leaves the datacenter, and it costs you nothing until you clear a terabyte in a single month. Almost no side project or early-stage app touches that ceiling.
There is no permanent free tier. When the 60-day trial ends, you start paying. But the entry price is low and the bandwidth inclusion makes the effective cost much lower than the sticker price implies.
Vultr offers $300 in trial credits for new accounts. Unlike DigitalOcean's 60-day window, Vultr does not publish a time limit on the trial credits — they run until exhausted. A small instance at $6 per month gives you substantial runway on $300, though Vultr enforces activity requirements to prevent dormant account abuse, so this is not a "sign up and ignore it" situation.
Post-trial pricing closely mirrors DigitalOcean. The Cloud Compute entry point is $6 per month for 1 GB RAM, 25 GB SSD, and 1 TB bandwidth. There is also a $2.50 per month tier: 512 MB RAM, 10 GB SSD, 500 GB bandwidth. That $2.50 instance is useful for very lightweight workloads, DNS-only services, or jump hosts — and there is no equivalent at DigitalOcean at that price point. Bandwidth is included on both tiers.
Railway operates on a completely different model. There is no large trial credit. New accounts get a 30-day trial with a one-time $5 credit, no credit card required. That is enough to explore the platform and run a small app for a month.
After the trial, the Hobby plan costs $5 per month flat, and that $5 plan fee includes $5 of usage credits. If your workload consumes $5 or less of compute in a month, your net cost above the plan fee is zero. A web service running on 256 MB RAM and 0.1 vCPU, active 24/7, costs approximately $2.90 per month at Railway's current rates (RAM: $10.14/GB/month, CPU: $20.31/vCPU/month). That fits entirely within the Hobby credit. Add a small Postgres database on similar specs and you might clear $5 slightly, but many side projects run indefinitely on Hobby without a single overage charge.
The Pro plan is $20 per month with $20 in credits. For teams or larger workloads, the credit model scales the same way.
| Provider | Trial Credit | Expiration | Post-Trial Entry | Bandwidth Included | Permanent Free Tier |
|---|---|---|---|---|---|
| AWS | Up to $200 | 6 months (hard) | Varies by service | 100 GB/mo free, then $0.09/GB | Limited (Lambda/DynamoDB) |
| DigitalOcean | $200 | 60 days (hard) | $6/mo (1 vCPU, 1 GB) | 1 TB/mo at $6/mo | None |
| Vultr | $300 | Until exhausted | $2.50/mo or $6/mo | 500 GB–1 TB/mo included | None |
| Railway | $5 (30 days) | 30 days | $5/mo Hobby (incl. $5 credit) | $0.05/GB egress | $5 credit resets monthly |
This is the section most comparison articles skip. The trial numbers are easy to find. What you pay at month three is the number that actually matters for a project with any legs.
On AWS, post-trial billing depends entirely on what you are running. There is no predictable floor. EC2, RDS, data transfer, NAT Gateway, CloudWatch logs, API Gateway calls — each service has its own meter. A small production app can easily run $50 to $150 per month on AWS without any obvious extravagance. The power and flexibility are real, but so is the complexity of the bill.
On DigitalOcean, a solo developer running a web app, a managed Postgres database, and occasional object storage usage can realistically keep things under $25 to $30 per month. The $6 Droplet with 1 TB bandwidth included handles most side project traffic without any overage. Managed Postgres starts at $15 per month for the smallest node. That is a complete stack for well under $25.
Vultr's post-trial pricing is nearly identical to DigitalOcean at the $6 tier. The $2.50 tier is a differentiator for lightweight workloads where 512 MB RAM is genuinely sufficient — there is no DigitalOcean equivalent at that price point.
Railway's ongoing cost model is the most unusual and the most underrated for small workloads. If you do not exceed $5 of compute, your Hobby plan costs $5 and that is the entire bill. Egress is charged at $0.05 per GB, so a traffic-heavy app will see that show up, but a typical API or internal tool with modest traffic may never trigger egress charges that matter.
The trial credit number is not the right metric. What matters is the post-trial floor: the minimum you can pay and still run a real service. On Vultr that floor is $2.50. On DigitalOcean and Railway it is effectively $5 to $6. On AWS it is difficult to define because the billing surface is too large to have a reliable floor for anything beyond the most minimal Lambda-only workload.
Railway's monthly credit reset is underappreciated. It is not a one-time trial. Every month, your $5 Hobby plan fee buys you $5 of compute. That is a recurring subsidy that makes small workloads genuinely free to operate above the plan cost. No other provider in this comparison does that.
DigitalOcean and Vultr's bandwidth inclusion is also frequently undersold. The moment you look at AWS's $0.09/GB egress rate and compare it to "1 TB included at $6/month," the true cost gap becomes obvious for any app with real traffic.
For a deeper look at what AWS credits actually cover and how fast they run out on different workload types, the AWS Free Tier Credit Calculator lets you model your specific setup before you commit. Punch in your intended instance size, storage, and data transfer, and it shows you how long your credits last and what the bill looks like after they expire.
If you want to compare post-trial pricing across all four providers on a specific workload, the Hosting Cost Calculator covers DigitalOcean, Vultr, AWS, and Linode with side-by-side monthly cost estimates based on your actual RAM, CPU, and storage needs.
The short version: Vultr wins on trial credit volume. Railway wins on long-term affordability for small workloads. DigitalOcean wins on predictable post-trial billing with no egress surprises. AWS wins when you need the AWS ecosystem and are willing to manage a more complex billing model to get it.