--- title: "Free compute for side projects: VPS, databases and AI" date: 2026-09-04 updated: 2026-09-04 description: "Choose free VPS hosting, serverless APIs, databases and GPU compute, with recurring allowances, billing limits and monthly changes kept separate." reviewed_edition: "free-compute-september-2026" draft: false --- Free compute can cover a small API, a database, an occasional GPU job or even a Linux server. The awkward part is that all four can be advertised as “free hosting” while giving you very different things. Start with what needs to run, then check what happens when the allowance runs out. This is the maintained shortlist from the monthly free-compute series. The provider documentation linked below was checked on 4 September 2026. These are documentation checks, not a claim to have opened and tested every account. Regional capacity, account eligibility and provider terms still matter. ## Which kind of free compute fits the job? - **A Linux machine you control:** look at Oracle Cloud Always Free. You get a VM, but finding capacity and keeping an idle instance are part of the bargain. - **A small web API or webhook:** Cloudflare Workers is a useful starting point when each request needs little CPU time. - **An intermittently used database:** Neon gives you Postgres that sleeps between requests; Aiven covers MySQL and several other managed services. - **A short agent task:** Vercel Sandbox provides isolated execution with a session limit. It is a different purchase decision from an always-on server. - **A burst of GPU work:** Modal's recurring credit can pay for jobs, rather than reserving a GPU that sits idle. - **Calling a hosted AI model:** Workers AI has a daily allocation. An inference allowance does not give you a GPU machine to train your own model on. A free plan that stops at a limit is easier to budget than a free allowance inside a billable account. A trial credit has an expiry or a finite balance. Shared capacity may be free and still unavailable when you want it. Keep those distinctions attached to the recommendation. ## Free VPS hosting: Oracle Cloud Always Free [Oracle's Always Free resource catalogue](https://docs.oracle.com/en-us/iaas/Content/FreeTier/freetier_topic-Always_Free_Resources.htm) includes up to two AMD micro VMs and an Ampere A1 allowance of 1,500 OCPU-hours and 9,000 GB-hours each month. Oracle describes the latter as equivalent to 2 OCPUs and 12 GB RAM. The tenancy also has 200 GB of combined boot and block storage. The catch is capacity in your home region. An advertised allocation is not a reservation, and Oracle can reclaim idle instances. Those constraints make it useful for experiments you can recover, but a poor basis for promising somebody an uninterrupted service for nothing. If you already own a spare machine, [my experience of making a home server useful](/posts/you-need-a-home-server/) covers the other route, including why self-hosting everything was not the answer. Electricity and maintenance still count. ## Small APIs: Cloudflare Workers [Workers Free](https://developers.cloudflare.com/workers/platform/limits/) includes 100,000 requests per day, 10 milliseconds of CPU time per HTTP request and 128 MB memory. Reaching the daily request limit produces errors until the quota resets at midnight UTC. That is a practical shape for a small API, webhook or bit of application glue. The CPU limit is the important qualification: a high request count does not turn it into a general-purpose VM. Storage and other Cloudflare products have their own allowances; check each resource the application actually uses. ## Free databases: Neon and Aiven [Neon's Free plan](https://neon.com/faqs/free-plan-limits-and-quotas) gives each project 100 CU-hours of compute per month, 0.5 GB storage and 5 GB public network transfer. Compute scales to zero after five minutes of inactivity. Exhausting compute or transfer suspends it until the next billing period or an upgrade; the Free plan does not bill overages. That makes sense for an intermittent Postgres application with a small dataset. It is less comfortable for something that must answer continuously, because the compute budget is finite and sleeping is part of the plan. [Aiven's free services](https://aiven.io/docs/platform/concepts/service-pricing) cover MySQL, PostgreSQL, Kafka, OpenSearch and Valkey without a credit card or a fixed trial period. Aiven can power down inactive services and change their region or configuration. Choose the actual database your project needs, then check its individual service limits. ## Agent sandboxes: Vercel Sandbox [Vercel Sandbox on Hobby](https://vercel.com/docs/sandbox/pricing) includes five active CPU-hours, 420 GB-hours of provisioned memory and 5,000 creations per month. A session can run for up to 45 minutes, with ten sandboxes running concurrently. Exceeding the included quota pauses new sandbox creation rather than charging Hobby overages. CPU-hours and memory-hours are different budgets. Waiting on an API can use little CPU while the allocated memory continues to count. Check both before assuming that a quiet agent can stay running all month. ## GPU jobs and hosted AI APIs [Modal's Starter plan](https://modal.com/pricing) includes $30 of compute credit each month. Its CPU, memory and GPU prices determine how much work that buys, so there is no honest single number of “free GPU hours” without specifying the hardware and job. It is a useful option to compare for burst workloads; it is not a permanent GPU reservation. [Modal requires a payment method](https://modal.com/docs/guide/billing) and can charge beyond the credit. Set an appropriate [Workspace spend limit](https://modal.com/docs/guide/budgets) before leaving jobs running. [Cloudflare Workers AI](https://developers.cloudflare.com/workers-ai/platform/pricing/) includes 10,000 Neurons per day. Neurons are Cloudflare's usage unit, and the amount of inference that buys depends on the model. On Workers Free, further operations fail after the allowance is exhausted. Workers Paid can charge for additional usage. Some models require a paid billing method or prepaid AI Gateway credits, so the daily allocation does not make the entire model catalogue available free. ## Before building around an allowance Check whether the plan permits your intended use, whether a card is required, and whether hitting a limit stops work or starts a bill. Then check the less exciting parts: disk, public IP addresses, outbound traffic, idle shutdown and data retention. Keep a backup or an export route for anything you would mind losing. The monthly editions below contain the broader catalogue and a dated account of changes and corrections. This page stays at the same address as the shortlist changes, so it is the useful one to bookmark or link to. ## Latest edition [Free compute in September 2026: changed limits and corrections](https://billiem.uk/posts/free-compute-september-2026/) - 4 September 2026 ## Monthly archive - [Free compute in September 2026: changed limits and corrections](https://billiem.uk/posts/free-compute-september-2026/) - September 2026 - [Free compute in August 2026: GPU notebooks and agent tools](https://billiem.uk/posts/free-compute-august-2026/) - August 2026 - [Free compute in July 2026: VPS limits, sandboxes and AI](https://billiem.uk/posts/free-compute-july-2026/) - July 2026 - [Free compute for side projects: June 2026 free-tier guide](https://billiem.uk/posts/free-compute-for-ai-side-projects/) - June 2026