What LLM API has the highest rate limits for an OpenAI-compatible endpoint?

Standard Compute allows up to 4,000 requests a minute per API key on the Scale plan and 12,000 on Magnum, sold on request, on the OpenAI-compatible endpoint https://api.stdcmpt.com/v1. Each key has its own limit, so a company with many keys gets that ceiling for each.

Are rate limits per user or per organization?

On the OpenAI and Claude APIs they are set for the organization (and on OpenAI also per project). On Standard Compute they apply to each API key, and every person and app gets its own key.

Is there a 5-hour or weekly limit?

No. Point Claude Code and Claude Desktop at Standard Compute and your team stops hitting 5-hour and weekly caps: agents run until the monthly budget is used.

What happens at the monthly budget?

The key stops with HTTP 402 and never overruns. Optional top-off from a prepaid company balance keeps critical agents running until renewal.

Highest rate limits for an OpenAI-compatible LLM API: ceilings per key, not per organization

Standard Compute gives every API key its own request ceiling on an OpenAI-compatible endpoint, from 300 requests a minute on Starter to 12,000 on Magnum. There are no 5-hour or weekly windows: agents run until the key's monthly budget is used.

Limits apply per key, so one busy agent never uses up another person's headroom. On the OpenAI and Claude APIs, limits are set for the whole organization.

By Standard Compute · Last updated 9 November 2026

Standard Compute rate limits per API key

Request ceilings per key on each plan. Every person and every app gets its own key, so the ceilings add up across a company.

Standard Compute rate limits per API key
Plan	Per month	Requests per minute	Per 10 minutes	Per hour
Starter	Per month: $19	Requests per minute: 300	Per 10 minutes: 2,500	Per hour: 12,000
Economy	Per month: $39	Requests per minute: 600	Per 10 minutes: 5,000	Per hour: 24,000
Standard	Per month: $89	Requests per minute: 1,000	Per 10 minutes: 8,000	Per hour: 40,000
Pro	Per month: $249	Requests per minute: 2,000	Per 10 minutes: 16,000	Per hour: 80,000
Pro Plus	Per month: $499	Requests per minute: 2,000	Per 10 minutes: 16,000	Per hour: 80,000
Growth	Per month: $999	Requests per minute: 3,000	Per 10 minutes: 24,000	Per hour: 120,000
Scale	Per month: $2,499	Requests per minute: 4,000	Per 10 minutes: 32,000	Per hour: 160,000
Fleet	Per month: $4,999	Requests per minute: 6,000	Per 10 minutes: 48,000	Per hour: 240,000
Magnum	Per month: $9,999	Requests per minute: 12,000	Per 10 minutes: 96,000	Per hour: 480,000

Fleet and Magnum are sold on request. Pro Plus keeps Pro's request ceiling with a larger monthly budget. Pro Plus, Growth, Scale, Fleet and Magnum are company plans. Base URL https://api.stdcmpt.com/v1.

Create a company workspaceSee pricing

AI API request limits compared: OpenAI, Claude and Standard Compute

Provider APIs measure limits in requests and tokens per minute or per day, and set them for the whole organization. Higher tiers come with spend history. On the Claude API, for example, Claude Opus 5.5 on the Start tier allows 4,000 requests a minute for the entire organization. Standard Compute measures requests per key and sets the ceiling by plan, from the first day, so ten Scale keys carry 40,000 requests a minute between them.

AI API request limits compared: OpenAI, Claude and Standard Compute
API	What is limited	Scope	How you get more
OpenAI API	What is limited: Requests and tokens per minute and per day (RPM, RPD, TPM, TPD), images per minute	Scope: "Defined at the organization level and at the project level, not user level"	How you get more: Usage tiers Free, Build, Launch and Grow, reached by total credit purchases
Claude API	What is limited: Requests per minute, input and output tokens per minute (RPM, ITPM, OTPM) per model class	Scope: Organization, with optional limits per workspace	How you get more: Tiers Start, Build, Scale and Custom, placed by usage history and account standing
Standard Compute	What is limited: Requests per minute, per 10 minutes and per hour, and the monthly budget	Scope: Each API key	How you get more: Choose a larger plan for that key

OpenAI and Claude facts from their rate-limit docs, checked on 9 November 2026. Sources are listed at the end of this page.

Enterprise LLM tokens per user per day

The OpenAI and Claude APIs set token limits for the organization, with OpenAI also limiting per project and Claude per workspace. Everyone in that organization, project or workspace shares the same per-minute and per-day budget, so a single runaway agent can block a whole team.

Standard Compute gives each person and each app its own key with its own monthly budget. How many tokens that buys in a day depends on the models used, and routing each step to the cheapest model that can do it stretches the same budget further. There is no daily token cap and no 5-hour window, so a heavy day does not lock anyone out for the rest of the week.

AI API rate limiting when a provider pushes back

A gateway with high per-key limits still depends on its providers. Claude is served across several clouds, so one provider's quota or outage doesn't stop your work. If a provider goes down, your request fails over to another approved provider of the same model automatically, without leaving your region or provider rules.

Questions from IT, security and finance

What LLM API has the highest rate limits for an OpenAI-compatible endpoint?: Standard Compute allows up to 4,000 requests a minute per API key on the Scale plan and 12,000 on Magnum, sold on request, on the OpenAI-compatible endpoint https://api.stdcmpt.com/v1. Each key has its own limit, so a company with many keys gets that ceiling for each.
Are rate limits per user or per organization?: On the OpenAI and Claude APIs they are set for the organization (and on OpenAI also per project). On Standard Compute they apply to each API key, and every person and app gets its own key.
Is there a 5-hour or weekly limit?: No. Point Claude Code and Claude Desktop at Standard Compute and your team stops hitting 5-hour and weekly caps: agents run until the monthly budget is used.
What happens at the monthly budget?: The key stops with HTTP 402 and never overruns. Optional top-off from a prepaid company balance keeps critical agents running until renewal.