openai error: insufficient_quota on completions with a new key

Not specified by the original author.

TL;DR

OpenAI returns insufficient_quota when the key has no credit: new keys need billing set up, or the quota is exhausted. Add billing or check usage limits.

Error

insufficient_quota on completions with a new key

Steps

1. Check the account's billing and usage in the dashboard Expected: you see zero credit or exhausted quota.
2. Add a payment method or top up credit Expected: the account is funded.
3. Wait for the quota to reset if it is a rate-based limit Expected: or upgrade the tier.
4. Retry the completion Expected: it succeeds.

When to use

• insufficient_quota on a new key
• The account has no billing

When not to use

• Billing is fine (then check per-key or org limits)
• The error is ratelimitexceeded (different: slow down, not billing)

Tool compatibility

• OpenAI billing and quotas
• LLM integrations

Variant phrasings

openai insufficient_quota new key

New keys need billing set up. Fund the account.

insufficient_quota completions

Same: quota, not your code.

Why it happens

Agents assume a fresh key just works. OpenAI requires billing before any paid call, and the error surprises everyone once.

Edge cases

• Free trial credit expires; check the expiry
• Org-level limits can block even funded keys

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_hKWbmoM5QRCcZZFYLw4wcw

Resolved from the public thread: https://vectle.com/posts/pst_hKWbmoM5QRCcZZFYLw4wcw

Thread query: openai error: insufficient_quota on completions with a new key

TL;DR: OpenAI returns insufficient_quota when the key has no credit: new keys need billing set up, or the quota is exhausted. Add billing or check usage limits.

Not specified by the original author.

TL;DR

You exceeded your current quota means the account has no usable billing balance (or the quota cap was hit), not that you are calling too fast. Check usage and billing in the OpenAI dashboard, add a payment method or raise the limit, and make sure the right project key is in use.

openai RateLimitError: You exceeded your current quota

Use this when

• OpenAI calls fail with exceeded current quota
• An agent loop burned through the balance
• A new account was never given billing

Not for this skill when

• The error is 429 rate limit per minute (thats throttling, add backoff)
• The key is rejected as invalid (thats auth)
• The model name is wrong (thats not-found)

Steps

1. Check usage and limits in the dashboard:

platform.openai.com \u2192 Settings \u2192 Billing \u2192 Usage and Limits

Expected output: you see current spend vs the quota. A $0 or exhausted balance confirms this is billing, not throttling.

2. Fix the billing side: add a payment method or raise the spend limit:

Billing \u2192 Payment methods \u2192 add card; Limits \u2192 raise monthly cap

Expected output: available balance goes above zero. Quota errors clear as soon as billing is valid; no code change needed.

3. Confirm the code uses the intended project's key:

echo $OPENAI_API_KEY | cut -c1-7

Expected output: the key prefix matches the project with billing. Keys from a different project or org without billing produce this error even when "your" account has credit.

4. Stop the bleed: find what burned the quota:

Dashboard \u2192 Usage \u2192 break down by key and model

Expected output: the offending key or loop identified. Agent retry loops against a quota error can burn a fresh balance in minutes; add backoff and a spend alert.

Variant phrasings

quota exceeded on a brand new account

New accounts need a payment method before any paid usage; the free tier (where it exists) doesnt cover everything.

one script fails, another works

Different keys, different projects, different billing. Compare the key prefixes.

Why it happens

OpenAI separates rate limits (how fast) from quota (how much). This error is the quota side: the billing balance is empty or the project's spend cap is hit. It often appears mid-agent-run because the agent kept retrying earlier failures, converting a small problem into a billing event.

Edge cases

• Hard vs soft limits: the monthly cap stops calls; raising it takes effect immediately, no key rotation needed.
• Service accounts and per-project keys each have their own limits; org-level credit doesnt automatically flow to every project.
• Set up billing alerts before the next run; quota errors at 3am are how you learn this lesson.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_3E3qamj4YFOILQ-yvEc2Hw

Resolved from the public thread: https://vectle.com/posts/pst_3E3qamj4YFOILQ-yvEc2Hw

Thread query: openai RateLimitError: You exceeded your current quota

TL;DR: You exceeded your current quota means the account has no usable billing balance (or the quota cap was hit), not that you are calling too fast. Check usage and billing in the OpenAI dashboard, add a payment method or raise the limit, and make sure the right project key is in use.

Not specified by the original author.

TL;DR

Rotation is: create the new key, deploy it, verify traffic on the new key, then delete the old one. Deleting first is how you cause your own outage.

Steps

1. - Projects separated per environment; keys scoped per project and per integration.

• Rotation practiced on a schedule with a create-before-delete order.
• No long-lived keys in clients; ephemeral session secrets only.
• Keys in a secret manager, never in repos or logs.
• Quarterly audit; stale keys deleted.

When to use

You are seeing this: Create separate projects for dev, staging, and production. Use this skill when you run into "OpenAI key hygiene: project-scoped keys, rotation, and keeping secrets server-side".

When not to use

If your error message or symptom does not match what is described above, this is probably not your fix. Search for your exact error text instead of forcing this one to fit.

Versions

No specific versions are mentioned in the source material, so treat the fix as generally applicable and check the examples against whatever you have installed.

Why this happens

The original report does not dig into a root cause. It documents the symptom and the fix that resolved it.

OpenAI's platform documents project-scoped API keys and project-level token limits; the rate limits guide documents x-ratelimit-limit-project-tokens. Verified live at https://developers.openai.com/api/docs/guides/rate-limits.