Anthropic API Error rate_limit_error: Rate limit exceeded (429)

Identify: does your log match?

anthropic.RateLimitError: Error code: 429 - {"type": "error", "error": {"type": "rate_limit_error", "message": "This request would exceed your organization's rate limit of 30,000 input tokens per minute."}, "request_id": "req_011CUQ7x2kGd8sWb1mHf4ZqA"}
retry-after: 12

1. High

   Input tokens per minute (ITPM)

   Large prompts (long documents, big tool definitions) use up the per-minute input token budget long before the request count limit.

2. Medium

   Bursty concurrency

   A batch job sends many requests in the same second; limits are enforced continuously, so short bursts trip them even when the minute average is fine.

3. Medium

   Low usage tier

   New organizations start on lower limits that rise with spend.

Fix

for (const doc of docs) await client.messages.create(build(doc));

import Anthropic from "@anthropic-ai/sdk";

for (const doc of docs) {
  try { await client.messages.create(build(doc)); }
  catch (e) {
    if (!(e instanceof Anthropic.RateLimitError)) throw e;
    const wait = Number(e.headers?.["retry-after"] ?? 10);
    await new Promise(r => setTimeout(r, wait * 1000));
  }
}

Verify: reproduce it on purpose

$ for i in $(seq 1 80); do curl -s -o /dev/null -w "%{http_code}\n" https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" -H "anthropic-version: 2023-06-01" -H "content-type: application/json" \
  -d '{"model":"claude-haiku-5-5","max_tokens":16,"messages":[{"role":"user","content":"hi"}]}' & done | sort | uniq -c

Prevent

• Read the anthropic-ratelimit-* response headers and slow down before you hit zero.
• Cap concurrency per model instead of firing all requests at once.
• Move non-urgent work to the Message Batches API.