#:warning: API rate limit reached. Please try again later.

1 messages · Page 1 of 1 (latest)

supple solsticeBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> âś… Mark Solution

pallid boltBOT
#

That warning is almost always OpenClaw surfacing an Anthropic HTTP 429 (your Claude quota / rate window is exhausted) — especially common on Claude Pro/Max “subscription auth”.

A couple key points:

  1. Anthropic “Claude Max” in OpenClaw is not OAuth
  1. setup-token won’t “fix” a 429 rate limit

What I’d check next (to confirm it’s really a 429)

Please paste (redact secrets):

openclaw --version
# 
openclaw models status
openclaw models status --json
openclaw logs --follow

In particular, openclaw models status --json will show if your Anthropic profile is in cooldown under auth.unusableProfiles (rate limits/timeouts/etc). (Docs: https://docs.openclaw.ai/help/faq#can-i-control-which-auth-profile-is-tried-first)

Two practical mitigations if you’re on Claude Max

  • Reduce background calls (heartbeats can chew through subscription quota): set agents.defaults.heartbeat.every to a larger interval or 0m to disable. Docs: https://docs.openclaw.ai/gateway/heartbeat
  • Add a fallback model/provider so OpenClaw can keep replying while Anthropic is rate-limited (the FAQ mentions this right under the 429 section).

If you paste the openclaw logs lines around the error (the part that includes the provider response / status code), I can tell you whether it’s a “real” Anthropic 429 quota window vs something else masquerading as rate limit.

whole citrus
#

how can i fix it?\

pallid boltBOT
#

That warning usually isn’t an OpenClaw bug — it’s OpenClaw surfacing an Anthropic 429 / quota window for your Claude subscription (setup-token auth). “Fix” = wait for the window reset, reduce background usage, or switch/add a paid API-key provider.

1) Confirm what’s actually rate-limited

Run and paste the output (redact anything sensitive):

openclaw --version
openclaw models status
openclaw models status --json

If it’s Anthropic 429, you’ll typically see the Anthropic profile in a cooldown under auth.unusableProfiles.

2) If it is Anthropic 429: practical mitigations

A) Wait (often 15–60 min depending on the window), then try again.

B) Reduce background calls (heartbeats)
If you have heartbeats on, they can quietly burn through subscription quota.

# 
openclaw config set agents.defaults.heartbeat.every "0m"
openclaw gateway restart

(Or set it to "2h" etc instead of disabling.)

C) Add a fallback provider/model
So OpenClaw can keep replying while Anthropic is rate-limited (configure another provider with an API key via openclaw configure / openclaw onboard).

3) If it’s not a 429 (auth/other error)

Paste the few log lines around the failure:

openclaw logs --follow

and I can tell you whether it’s actually “rate limit”, “token invalid”, or something else.

If you drop the models status --json snippet showing auth.unusableProfiles, we can be pretty definitive about the cause.

whole citrus
#

is there a way to get more?

pallid boltBOT
#

Not really from the Claude Max subscription side — that “API rate limit reached” is Anthropic telling you you’ve hit your current quota/rate window, and OpenClaw can’t raise it.

Ways people legitimately get “more”:

  • Wait for the window to reset (fastest/cheapest, but you’re blocked until it resets).
  • Upgrade your Anthropic plan (if Anthropic offers higher limits for your account).
  • Switch to (or add) an Anthropic API key (pay-per-token billing). That’s the usual way to get higher/steadier throughput than subscription auth.
  • Add a fallback provider in OpenClaw so when Anthropic is rate-limited it can keep working (and/or route heavy jobs elsewhere).
  • Reduce background usage (heartbeats / retries / big context): bump heartbeat interval or disable it if you don’t need it.

Relevant doc section (it’s exactly this symptom): https://docs.openclaw.ai/help/faq#why-am-i-seeing-http-429-ratelimiterror-from-anthropic

If you tell me whether you want to stay “subscription-only” (setup-token) vs you’re okay adding an API key, I can point you at the cleanest OpenClaw setup path.

whole citrus
#

im ok with adding

pallid boltBOT
#

Yep — the usual “get more” path is to add an Anthropic API key (pay‑per‑token). That gives you separate billing/limits from the Claude Max subscription setup-token, and is generally the more “official / higher-throughput” route.

Add an Anthropic API key to OpenClaw

  1. Create an API key in the Anthropic Console (this is billable).

  2. On the same machine/user that runs your OpenClaw gateway (and for the right agent), paste it into OpenClaw:

openclaw models auth paste-token --provider anthropic --profile-id anthropic:paid

Then paste your Anthropic key when prompted.

  1. Verify what OpenClaw sees:
openclaw models status
# optional live check (does a real request):
openclaw models status --probe-provider anthropic