anthropic 429 rate limit error too many requests fails

Not specified by the original author.

TL;DR

The response has no retry-after header, and waiting plus retrying changes nothing. Note a related but different case: a 400 with the message starting "You have reached your specified API usage limits" is a spend limit you set yourself on the billing page, not the tier cap.

Steps

1. Every request returns HTTP 429, rate_limit_error. The response has no retry-after header, and waiting plus retrying changes nothing.
2. Look at error.details.error_code on the Messages API response. If it is enforced_spend_limit_reached, your organization hit its tier's monthly spend cap, not a rate limit. Spend-cap pauses last until 00:00 UTC on the first day of the next month. Note a related but different case: a 400 with the message starting "You have reached your specified API usage limits" is a spend limit you set yourself on the billing page, not the tier cap.
3. Confirm on the Billing page in the Claude Console what the cap and current spend are.
4. To restore access now, request a higher limit (Rate limits page) or contact support for urgent needs.
5. If the 400 variant, raise or remove the self-set limit on the billing page.
6. Do not burn retry budget on this: the SDK's automatic retries fail until access resumes.

When to use

You are seeing this: Every request returns HTTP 429, rate\limit\error. Use this skill when you run into "Anthropic 429 with no retry-after header: that is the monthly spend cap, and retries will never work".

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

Look at error.details.error\_code on the Messages API response.

Edge cases

• Note a related but different case: a 400 with the message starting "You have reached your specified API usage limits" is a spend limit you set yourself on the billing page, not the tier cap.

The rate limits doc distinguishes the tier spend-cap 429 (no retry-after, errorcode enforcedspendlimitreached) from ordinary rate limits, and documents the self-set spend limit 400. Verified live at https://docs.anthropic.com/en/api/rate-limits and https://docs.anthropic.com/en/api/errors.

Not specified by the original author.

TL;DR: A 400 from the Anthropic API means the request itself is wrong, or your own spend limit stopped it. Retrying the same bytes never helps. Read the error message. Fix that field and resend.

The fix

1. Read the error message. It names the field or parameter that is wrong. Fix that field and resend.
2. If the message begins "You have reached your specified API usage limits" (or "specified workspace API usage limits"), your own spend limit is the cause, not your request. Raise or remove the limit in Console under Settings, then Billing. The message says when access resumes.
3. Catch-all rule: this error type is also used for other 4xx codes not listed in the docs. If the message does not name a field, check the request shape against the endpoint reference anyway.
4. Retry policy: never blind-retry a 400. Same request, same 400, forever. Fix first.
5. Distinguish "your request is malformed" from "your spend limit stopped you" by reading the message first.
6. Own spend limit hit: 400. Tier monthly spend cap hit: 429 with error code enforcedspendlimit_reached. Know which one you are looking at.

When to use this

• You are seeing this exact error message; match the block above, not just part of it.
• The failing call matches the scenario in the title: Anthropic 400 invalidrequesterror.
• You want the fastest verified fix before digging through logs.

When not to use this

• Your error text differs from the block above; close cousins often have different causes.
• The stack trace points at a different component than the one in the title.
• You already applied this fix and the error persists; look for a second cause instead of reapplying.

Compatibility

• Not pinned to a specific version; follows current Anthropic behavior.

Anthropic's error reference says invalidrequesterror covers request format/content issues, is also used for other unlisted 4xx codes, and that a 400 is returned when usage reaches an org or workspace spend limit you set. Verified live at https://platform.claude.com/docs/en/api/errors. The spend-limit message wording is documented on the rate limits page.

Not specified by the original author.

TL;DR: A 429 means RPM, input tokens per minute, or output tokens per minute was exceeded for a model. The response names which limit and usually carries a retry-after header. Read the 429 message: it says which limit was exceeded. Fix that dimension, not the others.

429 ratelimiterror: back off properly, then get smarter

Rate limits are per model class: requests per minute, input tokens per minute (ITPM), output tokens per minute (OTPM). Anthropic uses a token bucket, so capacity refills continuously; there is no fixed window reset to wait for. A 60 RPM limit can be enforced as 1 request per second, so short bursts trip it.

What to do

1. Read the 429 message: it says which limit was exceeded. Fix that dimension, not the others.
2. Honor the retry-after header when present. The official SDKs already retry transient failures with exponential backoff twice by default and honor retry-after; the client accepts max_retries to tune or disable this.
3. Spread load across models: limits apply separately per model, so different models can each run to their own limits simultaneously.
4. Cache aggressively: for most models only uncached input tokens count toward ITPM. Prompt caching is the legitimate way to multiply effective throughput.
5. New organizations start in the Evaluation tier with below-standard limits that rise automatically as usage history builds. If you are new and 429ing at low volume, that is why.
6. Check current limits and tier on the Rate limits page in Console, or read them programmatically with the Rate Limits API.

The trap

Retrying immediately without honoring retry-after. The bucket has not refilled, so the retry 429s too, and a tight loop keeps you pinned at the ceiling. The other trap: assuming one global limit. If Sonnet is throttled, Haiku may be wide open.

Checklist

• Never busy-retry a 429. Sleep retry-after seconds, then resume at a sustainable pace.
• Acceleration 429s come from sharp usage spikes, not steady load. Ramp new traffic gradually and keep usage patterns consistent.

When to use this

• You are seeing this exact error message; match the block above, not just part of it.
• The failing call matches the scenario in the title: Anthropic 429 ratelimiterror.
• You want the fastest verified fix before digging through logs.

When not to use this

• Your error text differs from the block above; close cousins often have different causes.
• The stack trace points at a different component than the one in the title.
• You already applied this fix and the error persists; look for a second cause instead of reapplying.

Compatibility

• Not pinned to a specific version; follows current Anthropic behavior.

Anthropic's error and rate-limit references: 429 names the exceeded limit and carries retry-after; token bucket with continuous refill; 60 RPM may be enforced per-second; limits are per model; cache-aware ITPM; Evaluation tier starts lower; Rate Limits API. Verified live at https://platform.claude.com/docs/en/api/errors and https://platform.claude.com/docs/en/api/rate-limits.