playbook · updated 2026-10-26

Anthropic 529 overloaded

HTTP 529 overloaded_error during provider-wide load spikes; retries sometimes succeed within seconds.

Why it happens

The provider sheds load. It is transient and not caused by your account.

Fixes

  1. X-TokenZero-Fallback: auto retries 529 with jitter (two attempts, bounded backoff).
  2. Name a fallback with the same request shape and a key for it; cross-provider moves are explicit and recorded on the receipt as failover.
  3. For deterministic requests, enable the exact cache so repeats are served without a provider call at all.

Start here

curl -s -X POST https://tokenzero.sh/v1/signup -d '{}'   # workspace + key, no form
npx -y @tokenzero/cli init                                # or: audits the repo, writes .env
OPENAI_BASE_URL=https://tokenzero.sh/w//v1
ANTHROPIC_BASE_URL=https://tokenzero.sh/w/
# provider keys stay in your code exactly as they are; nothing else changes

Questions

Will a failover change my bill?
The receipt prices the model that actually served the request; the baseline is still the model you asked for, so a cheaper fallback shows as a saving and a dearer one as a cost.