Pricing

Claude Code "API Error: 429": Rate Limit vs Usage Limit

Q: What is rate_limit_error in Claude Code?

It is the `error.type` / wording used in API and issue reports for 429 rate-limit responses, including spend-cap examples in Anthropic's rate-limits doc (verified Nov 7, 2026).

Q: Will waiting fix a 429?

Often for temporary limiting / `retry-after` cases. Spend-cap 429s without `retry-after` need the Console / workspace limit path instead of a short wait.

Claude Code says "API Error: 429" or rate_limit_error? How a 429 differs from usage limit reached, what causes it, and the documented wait-or-retry fixes.

By Engineer Foodie Desk Nov 7, 2026 ✓ Sources checked Nov 7, 2026

7 min read

Error strings quoted from Claude Code GitHub issues #69520, #87555, #97617 and #96872 (verified Nov 7, 2026). Causes and fixes checked Nov 7, 2026 in Claude Code’s error reference, Anthropic’s rate-limits docs and API error docs. Every source is listed at the bottom. No affiliate links on this page (how we work).

The exact errors: `API Error: Request rejected (429)`

The title string from #69520 (verified Nov 7, 2026):

API Error: Request rejected (429)

That issue’s body also repeats a longer line:

API Error: Request rejected (429) · Usage credits are required for long context requests.

Related rate-limit wording from #87555 and #97617 (verified Nov 7, 2026):

429 rate_limit_error

Anthropic’s current error reference documents the Request-rejected family like this (verified Nov 7, 2026):

API Error: Request rejected (429) · this may be a temporary capacity issue. If it persists, check https://status.claude.com.

And a separate, explicitly labeled twin:

API Error: Server is temporarily limiting requests (not your usage limit)

The one-line answer: a 429 is a request throttle, spend-cap refuse, or temporary capacity refuse — not the same string as Claude AI usage limit reached. Anthropic’s usage-limits intro in the error reference separates plan quota from this short-lived throttle family, and the “Server is temporarily limiting requests” section says it is “unrelated to your plan quota” and is retried automatically as of v2.1.199 (errors, verified Nov 7, 2026).

If your banner is the plan-quota wording, go to Claude Code usage limit reached. If it is 529 / overloaded, go to Claude Code 529 overloaded. This post owns API Error: 429 / rate_limit_error / Request rejected (429) only.

Picture I Love Lucy at the chocolate conveyor belt: the line is moving faster than the wrappers. That is a rate limit. It is not the same as the factory telling you your shift hours are used up for the week.

429 vs usage-limit-reached vs 529

Message	What it is	Sister post
Request rejected (429)	Rate / capacity / spend	You’re here
rate_limit_error	Same family	You’re here
Temp limiting (not usage)	Short throttle	You’re here
Usage limit reached	Plan session/week	Usage limit reached
529 / Overloaded	Server traffic jam	529 overloaded

Meanings from the Claude Code error reference (verified Nov 7, 2026) and the linked sister posts.

Causes reported in the issues and docs

Only what the sources actually say.

Repeated 429s during a session

#97617 (opened Sep 27, 2026, closed not_planned Oct 4, 2026 for needs-info/inactive) is the “429, token rate limited… AGAIN” report on Windows / VS Code. The bot closed it for missing info, so there is no maintainer root-cause write-up — only the repeated rate-limit symptom.

#96872 (opened Sep 24, 2026, still open) is more detailed: Max plan, ~6+ concurrent desktop sessions, plan usage around 26–27%, transcript with status: 429, error.type: rate_limit_error, message "Rate limited", rateLimits: null, retries up to maxRetries: 10. The reporter treats it as distinct from the five-hour/weekly quota. A commenter notes concurrency-sensitive bursts and says staggering launches helped — that is a user observation, not an official Anthropic rule. A published per-plan concurrent-session cap is not published on the rate-limits page fetched Nov 7, 2026.

429 while reported usage looks low

#87555 (opened Aug 18, 2026, closed not_planned Sep 20, 2026 as stale) is titled Persistent 429 rate_limit_error despite low usage (1%). The body describes usage at 1% with a persistent 429 rate_limit_error and a request id. No rate_limit_error JSON blob appears in that issue’s Errors array. Closed stale, no fix confirmed.

`Request rejected (429)` and the long-context credits line

#69520 (opened Jun 19, 2026, closed duplicate Jun 22, 2026 of #61692) titles API Error: Request rejected (429) and repeats API Error: Request rejected (429) · Usage credits are required for long context requests. Keep that issue wording distinct from the docs: Anthropic’s error reference has a separate entitlement string API Error: Usage credits required for 1M context · run /usage-credits… under its own section, not labeled as the Request-rejected-429 row (errors, verified Nov 7, 2026). Do not collapse those two into one cause.

What the official rate-limit docs add

Anthropic’s rate-limits doc: exceeding RPM / ITPM / OTPM returns 429; when retry-after is present, wait that many seconds. Spend-cap 429 example JSON uses "type":"rate_limit_error" with a message that the organization crossed its monthly API usage threshold, details.error_code: enforced_spend_limit_reached, and no retry-after. Acceleration limits and fast mode can also 429 with retry-after. A Claude Code workspace spend limit can return 429 with retry-after (versus an org self-set limit returning 400) (rate limits, verified Nov 7, 2026).

The API errors page matches that: 429 - rate_limit_error means the organization hit a rate limit, reached the usage tier’s monthly spend cap, or reached a spend limit on the Claude Code workspace; tier spend-cap 429 has no retry-after (API errors, verified Nov 7, 2026).

How to fix Claude Code API Error 429, in order

Each step says where it comes from.

Wait and retry with backoff. For the “Server is temporarily limiting requests (not your usage limit)” family, Anthropic says wait briefly; retries are automatic as of v2.1.199 (errors, verified Nov 7, 2026). When a retry-after header is present, the rate-limits doc says wait that many seconds.
Run /status and check plan usage separately. A stray ANTHROPIC_API_KEY can route you through a low-tier key (errors, verified Nov 7, 2026). If /status shows a real session or weekly plan cap, that is the usage-limit post, not this one.
Reduce concurrency / parallel sessions. The Request-rejected docs suggest reducing concurrency or trying a smaller model. #96872 reporters describe bursts across many desktop sessions; staggering is a user observation only.
Check provider console / rate-limit tier if you are on an API key. Use Anthropic’s rate-limits reference for RPM/ITPM/OTPM and spend-cap behavior. Workspace spend limits and org monthly thresholds are documented 429 paths.
Enable usage credits only when the official entitlement string says so. The 1M-context credits message is its own errors-doc section with /usage-credits. Do not assume every 429 wants credits — especially the “not your usage limit” throttle.
Check status.claude.com. The Request-rejected display string itself points there when the refuse persists. As of Nov 7, 2026 the page showed the aggregate banner All Systems Operational (verified Nov 7, 2026).
Still stuck? File /feedback with the full 429 line, whether retry-after appeared, /status output (redact secrets), and version.

What not to do: don’t farm accounts to dodge rate limits, don’t run a proxy “unlimiter,” and don’t treat a 429 as permission to smash the API harder. That is Groundhog Day without the character development — same morning, same refuse, worse odds.

FAQ

What does Claude Code API Error: 429 mean? A request throttle, temporary capacity refuse, or documented spend-cap / rate-limit refusal — Anthropic’s error reference and rate-limits docs cover the family (verified Nov 7, 2026).

Is API Error 429 the same as Claude AI usage limit reached? No. Plan session/weekly quota uses its own usage-limit messages; Anthropic even labels one 429 twin “Server is temporarily limiting requests (not your usage limit).”

What is rate_limit_error in Claude Code? It is the error.type / wording used in API and issue reports for 429 rate-limit responses, including spend-cap examples in Anthropic’s rate-limits doc (verified Nov 7, 2026).

Does API Error: Request rejected (429) mean I need usage credits? Not always. #69520 includes a long-context credits line in the issue body, but the docs keep a separate “Usage credits required for 1M context” entitlement string — do not collapse them.

Will waiting fix a 429? Often for temporary limiting / retry-after cases. Spend-cap 429s without retry-after need the Console / workspace limit path instead of a short wait.

Here’s what I’d do

Read the full line after 429. If it says the refuse is not your usage limit, wait, let retries finish, check status.claude.com. Run /status so a stray API key or a real plan-quota message does not hide in the same afternoon. Cut parallel sessions if you were stampeding the desktop app. Only chase usage credits when the entitlement string (or Console spend UI) says that is the gate.

If 429 storms keep wrecking deep-work blocks, keep a backup IDE warm. Our Claude Code alternatives list and AI coding tools compared page cover the options with dated prices. Plan-by-plan quota detail stays in Claude Code usage limits.

Plan names for older threads: Max is $100; Ultra is $200 (formerly Max 20x) (support 17728952, verified Nov 1, 2026).

Sources

Claude Code GitHub issue #69520 (API Error: Request rejected (429); closed duplicate): https://github.com/anthropics/claude-code/issues/69520 (verified 2026-11-07)
Claude Code GitHub issue #87555 (Persistent 429 rate_limit_error despite low usage (1%); closed stale): https://github.com/anthropics/claude-code/issues/87555 (verified 2026-11-07)
Claude Code GitHub issue #97617 (repeated 429 during session; closed needs-info): https://github.com/anthropics/claude-code/issues/97617 (verified 2026-11-07)
Claude Code GitHub issue #96872 (429 Rate limited with concurrent desktop sessions; open): https://github.com/anthropics/claude-code/issues/96872 (verified 2026-11-07)
Claude Code error reference (§ Request rejected (429); § Server is temporarily limiting requests; 1M-context credits sister section): https://code.claude.com/docs/en/errors (verified 2026-11-07)
Anthropic API rate limits (RPM/ITPM/OTPM; retry-after; spend-cap 429; workspace spend limit): https://platform.claude.com/docs/en/api/rate-limits (verified 2026-11-07)
Anthropic API errors (429 - rate_limit_error): https://docs.claude.com/en/api/errors (verified 2026-11-07)
Anthropic help, What is the Ultra plan (Max 20x renamed Ultra): https://support.claude.com/en/articles/17728952-what-is-the-ultra-plan (verified 2026-11-01)
Claude status page: https://status.claude.com (reachable 2026-11-07; aggregate “All Systems Operational”)

The exact errors: API Error: Request rejected (429)