Questions you probably have
The main points. If something else is missing, the API docs may have it.
Is there a catch?
No hidden fees. Plans renew until you cancel from the billing portal. Once access or credits are used, they stay used. Unused purchases can be reviewed under the Refund Policy.
Is this built for agents?
Yes. It is aimed at OpenClaw, Hermes, Pi, Conductor, OpenCode, Cursor, Claude Code CLI, Codex Desktop, and similar tools that need a stable API.
What models do I get?
Day and Week include the core set (GLM, MiniMax, DeepSeek, Sonnet, Haiku, Codex mini). Max adds Opus, the full GPT and Codex group, and GPT-5.6. Ultimate adds unlimited Fable 5 plus early access. Other plans include $200/day of Fable 5.
Can this get me banned?
No. The accounts used are enterprise accounts purchased directly by OpenLimits. Customer personal accounts are never involved.
Does it work with Claude Code CLI?
Yes. One variable and the usual features carry over. Most people are running in under 30 seconds.
What about rate limits?
No 5-hour or 7-day caps. A global 30 requests per second limit sits well above what normal sessions use.
Is setup actually 30 seconds?
Yes. It requires a key and one environment variable. The process has been timed on multiple occasions. Detailed steps are available on the How It Works page.
Can I track my usage?
Dashboard with live analytics, history, model splits, token counts, and a live feed.
What tools are supported?
Conductor, OpenCode, Claude Code CLI, Codex Desktop, OpenClaw, Hermes, Pi, Cursor, direct clients, and the Python and TypeScript SDKs.
Does it work with Codex Desktop?
Yes. Use the Codex/GPT group on plans that have it. Same key, same dashboard.
Billing questions
How does billing work?
Plans renew on their schedule until you stop them in the billing portal. Day every day, Week weekly, Max and Ultimate monthly.
What happens when I cancel?
Renewals stop. Access runs to the end of the paid period. Dashboard history stays. You can come back later.
Can I cancel anytime?
Yes. The portal stops future charges. You keep what you already paid for until the period ends.
Why is it so cheap compared to the API?
Accounts are shared across users. The cost per token drops because load is not constant for everyone.
Technical questions
Is my data private?
We do not store prompts or completions; they pass through without being logged. Only metadata such as the model, token counts, and timestamps is recorded for your analytics dashboard. All access to the models uses enterprise accounts purchased directly from the providers.
What's the latency overhead?
Edge nodes add less than 50 ms in most cases. For streaming you will not see a difference in how fast tokens arrive.
Can I use it for production apps?
Yes. Same models and reliability as the direct APIs, plus failover when one account is busy.
Do you support tool/function calling?
Yes. Tool use is supported in the same manner as the direct provider APIs and functions correctly with streaming responses.
Do you support PDFs and images?
Yes. Send media the same way you would to the direct API.
What if Anthropic or OpenAI is down?
If the upstream provider is unavailable, requests routed to that provider cannot complete. OpenLimits operates as a proxy and forwards traffic to the actual model providers. Automatic failover handles many transient errors (such as 429 and 529 responses) by switching to another account in the pool.
Still have questions?
A Day Pass gives the higher tier for a short test or heavy day. Cancel before it rolls if you want. See the plans.
Get Your API Key →