How Standard Compute keeps unlimited AI compute fair and reliable for every customer on the platform.
Standard Compute offers unlimited AI compute so you can build automations without worrying about per-token billing. "Unlimited" means there is no hard cap on the number of requests or tokens you can consume within your plan — but it does not mean unrestricted or unmonitored usage.
Our fair use policy exists to ensure that every customer receives consistent, low-latency access to the API. It prevents a small number of accounts from consuming a disproportionate share of shared infrastructure and protects the quality of service for the entire platform.
Each plan is designed around a typical usage profile. The Starter plan is intended for individual builders running lightweight automations — typically up to around 500 requests per hour with standard-complexity prompts. The Pro plan supports heavier production workloads with higher sustained throughput. The Enterprise plan includes dedicated capacity and custom limits agreed upon during onboarding.
These are guidelines, not rigid ceilings. Occasional bursts above guideline levels are expected and accommodated. We only intervene when sustained usage patterns significantly exceed the plan profile over a rolling 24-hour period.
If your usage consistently exceeds the guideline for your plan, we apply a graduated response. First, requests that exceed the sustained threshold are throttled — they still succeed but may experience higher latency as they are queued behind normal-priority traffic.
If elevated usage continues for more than 24 hours, we will send you an email notification explaining the situation and suggesting a plan upgrade that better fits your workload. Your existing requests continue to be served during this period, albeit with potential throttling.
We will never silently drop requests or suspend your account without prior notice. Our goal is to work with you to find the right plan, not to penalise growth.
Certain activities are strictly prohibited regardless of plan level. These include using the API as a general-purpose compute platform for tasks unrelated to AI inference (such as cryptocurrency mining or hash generation), using the API to launch or relay denial-of-service attacks against any target, and systematic web scraping or data harvesting where the LLM is used solely to extract and repackage third-party content at scale.
Attempting to reverse-engineer, probe, or benchmark upstream model providers through our API in ways that violate their terms of service is also prohibited. Generating content that violates applicable law — including CSAM, targeted harassment, or fraud — will result in immediate account termination and may be reported to the relevant authorities.
If you are unsure whether a particular use case falls within acceptable use, contact us at contact@standardcompute.com before deploying it.
We monitor aggregate request volumes, token throughput, and error rates at the account level. Monitoring is automated and focuses on quantitative patterns — we do not read, log, or analyse the content of your prompts or model responses.
Enforcement decisions are made by the operations team, not automated systems. A human reviews the usage data and account context before any throttling or outreach occurs. You will always receive an explanation and an opportunity to respond.
We may update this fair use policy as the platform evolves. Material changes — such as adjustments to guideline thresholds or new prohibited categories — will be communicated to all active accounts via email at least 14 days before they take effect.
If you have questions about this policy or want to discuss a specific use case, reach out to contact@standardcompute.com.