Loading...
The messages that tell you to slow down, one that is written but never appears — and what each says about your credits.
A rate limit is a ceiling on how many times something may be asked for in an hour. It is not your prompt being refused, not a fault in your project, and not a punishment — it is the service saying not this many, not this fast.
The part worth trusting is the same every time: the request was stopped before the expensive work started, so nothing came out of your balance. The Video Studio and AI Chat messages say so in as many words. The short one in Image Chat does not, but the same holds there.
Starting a video is capped at 8 generations per hour, per account. Ask for a ninth inside the same hour and a message appears, worded one of three ways depending on what the service could tell the page:
The time in the first two is the real reset moment, not a guess. Nothing is queued behind it: the clip never started, so when the hour frees up you press generate again yourself.
The cost panel re-checks the price every time you change model, length or resolution, so price checks have their own allowance of 240 an hour and cannot eat into your 8 generations. Watching a clip finish has its own allowance too — 720 an hour, which is a check every five seconds without stopping — and saving a finished clip has 30 an hour.
In practice: if you are told the video service has hit its limit, it is because eight clips were started, not because you were browsing prices.
When the model provider itself pushes back, the answer names it:
"Claude (Anthropic) is rate-limiting requests right now. Switch to another model or try again in a moment. Your tokens were not charged."
The same sentence is written for OpenAI and Gemini (Google), and reads The selected provider when the provider cannot be named.
This one is not about your account at all — it is VyroAI's own access to that model being throttled. So the fix in the message is the real fix: pick another model in the picker and send the message again straight away.
One message looks almost the same and is not a rate limit: "Claude (Anthropic) is unavailable — the provider account is out of credit. Switch to another model, contact the developer, or try again later. Your tokens were not charged."
Waiting will not clear that one. Switch model, and if every model says it, tell support — see Getting Help.
The Chat mode of the Image Creator has its own cap: 100 messages an hour, per account. Over that, a short message appears — "Rate limited. Try again shortly." — and the assistant does not answer. The message was stopped before any AI ran, so nothing came out of your balance. It does not say when the cap frees up, so wait a few minutes and send again.
This cap is separate from the Video Studio's eight an hour and from the model provider's own limits above.
The sign-in page is written to be able to say "Too many failed attempts. Please try again in N minutes." You cannot be shown it today. The check in front of it was moved to the server, and the client no longer stops anyone, so a wrong password keeps returning "Invalid email or password. Please try again." however many times you try.
Sign-in attempts are still counted on the server, but that counting never blocks the sign-in and its refusals are never shown to you. If a sign-in is refused for a reason other than the password, it is one of the blocks described in Sign-In Problems — a blocklisted address or a country restriction — not a rate limit.
Asking for a password reset too often does get refused, with "Too many reset attempts. Please wait a minute and try again." A minute is the whole of it. Resetting Your Password covers the rest.
None of these is a reason to buy credits or to open a support ticket on the spot. Wait out the clock the message gives you, or switch model. If it keeps happening, /status says whether it is us — see The Status Page.
8 of 15 in Support & Legal · Last checked 11 September 2026 · Previous: When a Page Won't Load