LLM request rejected: You're out of extra usage. Add more at claude.ai/settin | Friends of the Crustacean 🦞🤝 | Page 1

manic compass Apr 20, 2026, 10:23 PM

#

I re-uped my extra use balance and it shows in my account but i am still getting this error via discord. what do i need to do?

next cedarBOT Apr 20, 2026, 10:23 PM

#

rain loomBOT Apr 20, 2026, 10:26 PM

#

That’s usually not Discord, it’s the Anthropic auth/billing route OpenClaw is using.

Most likely it’s one of these:

OpenClaw is using an Anthropic API key, not your Claude account
In that case, topping up at claude.ai/settings/usage won’t help. You’d need billing/credits on the Anthropic Console side instead.
OpenClaw cached the earlier billing failure
After a billing-style error, OpenClaw can temporarily mark the Anthropic profile as disabled for a while.

Can you paste these?

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

#

In the JSON, check auth.unusableProfiles for an anthropic:* entry with disabledUntil.

If you want to try the likely fix first:

openclaw models auth login --provider anthropic
openclaw gateway restart

Also, if the exact error says:

Extra usage is required for long context requests

then paste this too:

#

openclaw config get agents.defaults.models

That usually means context1m is enabled and the current Anthropic credential is not eligible for that route.

Docs:

manic compass Apr 20, 2026, 10:34 PM

#

🦞 OpenClaw 2026.4.5 (3e72c03) — I don't have opinions about tabs vs spaces. I have opinions about everything else.

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : anthropic/claude-sonnet-4-6
Fallbacks (0) : -
Image model : -
Image fallbacks (0): -
Aliases (1) : sonnet -> anthropic/claude-sonnet-4-6
Configured models (1): anthropic/claude-sonnet-4-6

Auth overview
Auth store : ~/.openclaw/agents/main/agent/auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (2): anthropic (1), openai-codex (1)

anthropic effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=1, api_key=0) | anthropic:manual=token:sk-ant-o...TuxLWAAA
brave effective=env:BSAsIr3F...C4SAL_cT | env=BSAsIr3F...C4SAL_cT | source=env: BRAVE_API_KEY
openai effective=env:sk-proj-...FrAoV2wA | env=sk-proj-...FrAoV2wA | source=env: OPENAI_API_KEY
openai-codex effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=1, token=0, api_key=0) | openai-codex:default=OAuth
openrouter effective=env:sk-or-v1...5d8e2977 | env=sk-or-v1...5d8e2977 | source=env: OPENROUTER_API_KEY
perplexity effective=env:sk-or-v1...5d8e2977 | env=sk-or-v1...5d8e2977 | source=env: OPENROUTER_API_KEY

OAuth/token status

anthropic
- anthropic:manual static
openai-codex usage: 5h 100% left ⏱4h 59m · Week 100% left ⏱6d 23h
- openai-codex:default ok expires in 0m

rain loomBOT Apr 20, 2026, 10:36 PM

#

Yep, this narrows it down.

You’re not on Anthropic API-key billing here. You’re using an Anthropic setup-token:

anthropic:manual=token

So your claude.ai/settings/usage top-up is the relevant bucket for this setup.

That means the likely problem is now one of these:

OpenClaw cached the earlier billing failure and still has that profile temporarily disabled
The saved Anthropic setup-token is stale and needs to be refreshed

Can you paste this next?

#

openclaw models status --json

and especially the auth.unusableProfiles part.

For your current auth path, the right thing to try is:

openclaw models auth setup-token --provider anthropic
openclaw gateway restart

Small correction to my earlier message: for your Anthropic setup, use setup-token, not auth login.

Also, I notice:

#

Fallbacks (0)

So if Anthropic is blocked, Discord just hard-fails instead of falling back to OpenAI/Codex.

Docs:

If you paste the --json, I can tell you immediately whether it’s a cached disabledUntil window or a token refresh issue.

manic compass Apr 20, 2026, 10:37 PM

#

sorry what? what do i need to do?

rain loomBOT Apr 20, 2026, 10:37 PM

#

No worries. Do these 2 things:

openclaw models auth setup-token --provider anthropic
openclaw gateway restart

Then try again in Discord.

Why: your Claude balance is probably fine now, but OpenClaw is still using the old Anthropic token/session from before the top-up.

If it still fails after that, paste this:

openclaw models status --json

#

and I’ll tell you the next step.

manic compass Apr 20, 2026, 10:38 PM

#

its asking for my anthropic set up token

#

its extra use so its supposed ot be under my login not api

rain loomBOT Apr 20, 2026, 10:39 PM

#

Yep, that’s expected, and don’t paste the token here in Discord.

Do this:

In a local terminal, run:
```
claude setup-token
```
Copy the token it gives you
It should start with:
```
sk-ant-oat01-
```
Paste that token into the OpenClaw prompt that’s asking for it

#

Then restart OpenClaw:
```
openclaw gateway restart
```
Try Discord again

If you don’t have claude installed/logged in, say that, and I’ll give you the other path.

#

Yep, exactly. You’re right.

This is your login-backed Anthropic route, not API-key billing.
The problem is likely just that OpenClaw is still using an old login token / old billing state from before you topped up.

So the fix is still:

Run:
```
claude setup-token
```
Paste that fresh token into the OpenClaw prompt
Restart:
```
openclaw gateway restart
```

#

That refreshes OpenClaw’s copy of your login auth so it sees the new extra-usage balance.

And again, don’t paste the token in Discord.

manic compass Apr 20, 2026, 10:42 PM

#

still getting same error

rain loomBOT Apr 20, 2026, 10:43 PM

#

Got it. Then we need the actual lockout/error state, not guesses.

Please run these and paste the output:

openclaw models status --json

Then, in another terminal:

openclaw logs --follow

Trigger the Discord error once, then paste the new log lines that appear.

What I’m looking for:

#

auth.unusableProfiles
disabledUntil
any Anthropic billing / extra usage line in the logs

That’ll tell us whether this is:

OpenClaw still holding a cached billing disable, or
Anthropic still rejecting the login token even after the top-up

Don’t paste any secrets, just the output.

manic compass Apr 20, 2026, 10:44 PM

#

{
"configPath": "/Users/butters/.openclaw/openclaw.json",
"agentDir": "/Users/butters/.openclaw/agents/main/agent",
"defaultModel": "anthropic/claude-sonnet-4-6",
"resolvedDefault": "anthropic/claude-sonnet-4-6",
"fallbacks": [],
"imageModel": null,
"imageFallbacks": [],
"aliases": {
"sonnet": "anthropic/claude-sonnet-4-6"
},
"allowed": [
"anthropic/claude-sonnet-4-6"
],
"auth": {
"storePath": "/Users/butters/.openclaw/agents/main/agent/auth-profiles.json",
"shellEnvFallback": {
"enabled": false,
"appliedKeys": []
},
"providersWithOAuth": [
"anthropic (2)",
"openai-codex (1)"
],
"missingProvidersInUse": [],
"providers": [
{
openclawer": "anthropic",
"effective": {
"kind": "profiles",
"detail": "~/.openclaw/agents/main/agent/auth-profiles.json"
},
"profiles": {
"count": 2,
"oauth": 0,
"token": 2,
"apiKey": 0,
"labels": [
"anthropic:manual=token:sk-ant-o...TuxLWAAA",
"anthropic:default=token:sk-ant-o...yqxLuwAA"
]
}
},
{
"provider": "brave",
"effective": {
"kind": "env",
"detail": "BSAsIr3F...C4SAL_cT"
},
"profiles": {
"count": 0,
"oauth": 0,
"token": 0,
"apiKey": 0,
"labels": []
},
"env": {
"value": "BSAsIr3F...C4SAL_cT",
"source": "env: BRAVE_API_KEY"
}
},
{
"provider": "openai",
"effective": {
"kind": "env",
"detail": "sk-proj-...FrAoV2wA"
},
"profiles": {
"count": 0,
"oauth": 0,
"token": 0,
"apiKey": 0,
"labels": []
},
"env": {
"value": "sk-proj-...FrAoV2wA",
"source": "env: OPENAI_API_KEY"
}
},
{
"provider": "openai-codex",
"effective": {
"kind": "profiles",
"detail": "~/.openclaw/agents/main/agent/auth-profiles.json"
},
"profiles": {
"count": 1,
"oauth": 1,
"token": 0,
"apiKey": 0,
"labels": [
"openai-codex:default=OAuth"
]
}
},
{
"provider": "openrouter",
"effective": {
"kind": "env",
"detail": "sk-or-v1...5d8e2977"
},
"profiles": {
"count": 0,
"oauth": 0,
"token": 0,
"apiKey": 0,
"labels": []
},
"env": {
"value": "sk-or-v1...5d8e2977",
"source": "env: OPENROUTER_API_KEY"
}
},
{
"provider": "perplexity",
"effective": {
"kind": "env",
"detail": "sk-or-v1...5d8e2977"
},
"profiles": {
"count": 0,
"oauth": 0,
"token": 0,
"apiKey": 0,
"labels": []
},
"env": {
"value": "sk-or-v1...5d8e2977",
"source": "env: OPENROUTER_API_KEY"
}
}
],

#

"unusableProfiles": [],
"oauth": {
"warnAfterMs": 86400000,
"profiles": [
{
"profileId": "anthropic:default",
"provider": "anthropic",
"type": "token",
"status": "static",
"source": "store",
"label": "anthropic:default"
},
{
"profileId": "anthropic:manual",
"provider": "anthropic",
"type": "token",
"status": "static",
"source": "store",
"label": "anthropic:manual"
},
{
"profileId": "openai-codex:default",
"provider": "openai-codex",
"type": "oauth",
"status": "ok",
"expiresAt": 1777588422217,
"remainingMs": 863412018,
"source": "store",
"label": "openai-codex:default"
}
],
"providers": [
{
"provider": "anthropic",
"status": "ok",
"profiles": [
{
"profileId": "anthropic:default",
"provider": "anthropic",
"type": "token",
"status": "static",
"source": "store",
"label": "anthropic:default"
},
{
"profileId": "anthropic:manual",
"provider": "anthropic",
"type": "token",
"status": "static",
"source": "store",
"label": "anthropic:manual"
}
]
},
{
"provider": "openai-codex",
"status": "ok",
"profiles": [
{
"profileId": "openai-codex:default",
"provider": "openai-codex",
"type": "oauth",
"status": "ok",
"expiresAt": 1777588422217,
"remainingMs": 863412018,
"source": "store",
"label": "openai-codex:default"
}
],
"expiresAt": 1777588422217,
"remainingMs": 863412018
}
]
}
}
}

rain loomBOT Apr 20, 2026, 10:45 PM

#

Ah, I think I see it.

You now have two Anthropic token profiles:

anthropic:manual
anthropic:default

So OpenClaw is probably still trying the old one first.

Force it to use the fresh one:

openclaw models auth order set --provider anthropic anthropic:default
openclaw gateway restart

Then try Discord again.

#

If that fixes it, great. If not, paste these next:

openclaw models auth order get --provider anthropic
openclaw logs --follow

right after reproducing the error once.

Docs: https://docs.openclaw.ai/gateway/authentication

#

Perfect, that rules out the cached-lockout theory.

"unusableProfiles": [] means OpenClaw is not currently holding your Anthropic profile in a disabled/cooldown state.

So the next step is a live Anthropic probe to see which token Anthropic is actually accepting:

openclaw models status --probe --probe-provider anthropic --probe-profile anthropic:default --probe-profile anthropic:manual

That makes a real check against Anthropic and should tell us one of these:

anthropic:default works, anthropic:manual fails
both fail
both work

If only one works, we force OpenClaw to use that one.

#

If both fail, then Anthropic is still rejecting the login token itself, even after the top-up.

Paste that output and I’ll tell you the exact fix.

manic compass Apr 20, 2026, 10:47 PM

#

ok i tried this and it didnt work

openclaw models auth order set --provider anthropic anthropic:default
openclaw gateway restart

rain loomBOT Apr 20, 2026, 10:47 PM

#

Right, that only forces which token OpenClaw uses. It doesn’t tell us whether Anthropic is accepting that token.

Please run this exact command next and paste the output:

openclaw models status --probe --probe-provider anthropic --probe-profile anthropic:default --probe-profile anthropic:manual

That’s the check that will show whether:

the new token works
the old token works
or Anthropic is still rejecting both

Once you paste that, I can give you the exact next step.

manic compass Apr 20, 2026, 10:48 PM

#

part 1/2

🦞 OpenClaw 2026.4.5 (3e72c03) — I don't judge, but your missing API keys are absolutely judging you.

│
15:47:57-07:00 [plugins] [claude-mem] OpenClaw plugin loaded — v1.0.0 (worker: 127.0.0.1:37777)
15:47:57-07:00 [plugins] [claude-mem] OpenClaw plugin loaded — v1.0.0 (worker: 127.0.0.1:37777)
15:47:57-07:00 [plugins] [claude-mem] OpenClaw plugin loaded — v1.0.0 (worker: 127.0.0.1:37777)
15:47:57-07:00 [plugins] [claude-mem] OpenClaw plugin loaded — v1.0.0 (worker: 127.0.0.1:37777)
15:47:57-07:00 [plugins] [claude-mem] Context injected via system prompt for agent=main
◇
Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : anthropic/claude-sonnet-4-6
Fallbacks (0) : -
Image model : -
Image fallbacks (0): -
Aliases (1) : sonnet -> anthropic/claude-sonnet-4-6
Configured models (1): anthropic/claude-sonnet-4-6

Auth overview
Auth store : ~/.openclaw/agents/main/agent/auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (2): anthropic (2), openai-codex (1)

anthropic effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=2 (oauth=0, token=2, api_key=0) | anthropic:manual=token:sk-ant-o...TuxLWAAA, anthropic:default=token:sk-ant-o...yqxLuwAA
brave effective=env:BSAsIr3F...C4SAL_cT | env=BSAsIr3F...C4SAL_cT | source=env: BRAVE_API_KEY
openai effective=env:sk-proj-...FrAoV2wA | env=sk-proj-...FrAoV2wA | source=env: OPENAI_API_KEY
openai-codex effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=1, token=0, api_key=0) | openai-codex:default=OAuth
openrouter effective=env:sk-or-v1...5d8e2977 | env=sk-or-v1...5d8e2977 | source=env: OPENROUTER_API_KEY
perplexity effective=env:sk-or-v1...5d8e2977 | env=sk-or-v1...5d8e2977 | source=env: OPENROUTER_API_KEY

#

part 2/2

OAuth/token status

anthropic
- anthropic:default static
- anthropic:manual static
openai-codex usage: 5h 100% left ⏱4h 59m · Week 100% left ⏱6d 23h
- openai-codex:default ok expires in 10d

Auth probes
┌─────────────────────────────┬───────────────────────────┬─────────────────────────────────────────────────────────┐
│ Model │ Profile │ Status │
├─────────────────────────────┼───────────────────────────┼─────────────────────────────────────────────────────────┤
│ anthropic/claude-sonnet-4-6 │ anthropic:default (token) │ ok · 1.8s │
│ anthropic/claude-sonnet-4-6 │ anthropic:manual (token) │ unknown · - │
│ │ │ ↳ Excluded by auth.order for this provider. │
└─────────────────────────────┴───────────────────────────┴─────────────────────────────────────────────────────────┘
Probed 1 target in 6.2s
butters@mac ~ %

rain loomBOT Apr 20, 2026, 10:48 PM

#

Got part 1. I need part 2, the part with the actual probe result/status for:

anthropic:default
anthropic:manual

That’s the bit that will say ok, billing, auth, etc.

#

Good, that’s actually very helpful.

It means your Anthropic token is working:

anthropic:default (token)  ok

So this is not a bad-token / bad-login problem anymore.

That means the remaining issue is on the actual Discord run path, not auth.
Next step:

In a terminal, run:
```
openclaw logs --follow
```

#

In Discord, send a fresh test message like:
```
hi
```
Paste the new log lines that appear right when it fails, especially anything mentioning:
- anthropic
- billing
- extra usage
- rate_limit

At this point we’ve narrowed it down to:

a Discord/runtime path issue, or
the actual request payload being rejected even though the token itself probes fine

We’re close now.

manic compass Apr 20, 2026, 10:50 PM

#

Log tail truncated (increase --max-bytes).
22:50:03+00:00 info plugins {"subsystem":"plugins"} [claude-mem] Context injected via system prompt for agent=main
22:50:03+00:00 warn agent/embedded {"subsystem":"agent/embedded"} {"event":"embedded_run_agent_end","tags":["error_handling","lifecycle","agent_end","assistant_error"],"runId":"c5580839-deb7-4de5-8131-e74c552f37f0","isError":true,"error":"LLM request rejected: You're out of extra usage. Add more at claude.ai/settings/usage and keep going.","failoverReason":"format","model":"claude-sonnet-4-6","provider":"anthropic","rawErrorPreview":"400 {"type":"error","error":{"type":"invalid_request_error","message":"You're out of extra usage. Add more at claude.ai/settings/usage and keep going."},"request_id":"sha256:945fa5a18929"}","rawErrorHash":"sha256:eb3c0237bd99","rawErrorFingerprint":"sha256:26d210060f19","httpCode":"400","providerErrorType":"invalid_request_error","providerErrorMessagePreview":"You're out of extra usage. Add more at claude.ai/settings/usage and keep going.","requestIdHash":"sha256:945fa5a18929"} embedded run agent end
22:50:03+00:00 warn agent/embedded {"subsystem":"agent/embedded"} {"event":"auth_profile_failure_state_updated","tags":["error_handling","auth_profiles","cooldown"],"runId":"c5580839-deb7-4de5-8131-e74c552f37f0","profileId":"sha256:154a23a3efe6","provider":"anthropic","reason":"format","windowType":"cooldown","windowReused":false,"previousErrorCount":0,"errorCount":1,"cooldownUntil":1776725433639,"failureCounts":{"format":1}} auth profile failure state updated
22:50:03+00:00 warn agent/embedded {"subsystem":"agent/embedded"} {"event":"embedded_run_failover_decision","tags":["error_handling","failover","assistant","surface_error"],"runId":"c5580839-deb7-4de5-8131-e74c552f37f0","stage":"assistant","decision":"surface_error","failoverReason":"format","profileFailureReason":"format","provider":"anthropic","model":"claude-sonnet-4-6","profileId":"sha256:154a23a3efe6","fallbackConfigured":false,"timedOut":false,"aborted":false,"rawErrorPreview":"400 {"type":"error","error":{"type":"invalid_request_error","message":"You're out of extra usage. Add more at claude.ai/settings/usage and keep going."},"request_id":"sha256:945fa5a18929"}","rawErrorHash":"sha256:eb3c0237bd99","rawErrorFingerprint":"sha256:26d210060f19","httpCode":"400","providerErrorType":"invalid_request_error","providerErrorMessagePreview":"You're out of extra usage. Add more at claude.ai/settings/usage and keep going.","requestIdHash":"sha256:945fa5a18929"} embedded run failover decision

rain loomBOT Apr 20, 2026, 10:51 PM

#

Okay, we found it.

OpenClaw is not stuck locally anymore. Anthropic itself is returning:

400 invalid_request_error: You're out of extra usage...

So the token is valid enough to probe, but Anthropic is still rejecting the real generation request.

Next check, on the same machine run:

claude -p "hi"

Then:

#

If claude -p also fails with extra usage, this is upstream on Anthropic’s side, not Discord/OpenClaw. Your top-up likely hasn’t propagated for real model calls yet.
If claude -p works, switch OpenClaw to the Claude CLI auth path instead of setup-token:

openclaw onboard

and choose Anthropic Claude CLI, then restart:

openclaw gateway restart

One more thing: this log line is probably an OpenClaw bug:

#

"failoverReason":"format"

That error is clearly billing-ish, not format. If you want to report that part:
https://github.com/openclaw/openclaw/issues

manic compass Apr 20, 2026, 11:21 PM

#

sorry what do i need to do next?

rain loomBOT Apr 20, 2026, 11:21 PM

#

Totally fair. Do this next:

claude -p "hi"

Then:

If it works: switch OpenClaw to Anthropic Claude CLI auth
```
openclaw onboard
```
choose Anthropic Claude CLI, then:
```
openclaw gateway restart
```

#

If it fails with the same extra-usage error: the problem is on Anthropic’s side, not Discord/OpenClaw.

If you want, just paste the output of claude -p "hi" and I’ll tell you which branch you’re in.

manic compass Apr 20, 2026, 11:30 PM

#

paste that into terminal?

rain loomBOT Apr 20, 2026, 11:30 PM

#

Yes, exactly, paste this into your local terminal, not Discord:

claude -p "hi"

Then tell me what it prints.

manic compass Apr 20, 2026, 11:31 PM

#

nothing

rain loomBOT Apr 20, 2026, 11:32 PM

#

Okay, two quick checks:

Run this:
```
claude --version
```
Tell me which happened with:
```
claude -p "hi"
```
- it returned straight back to the prompt
- it hung / kept waiting
- it printed an error

Paste whatever you see.

manic compass Apr 20, 2026, 11:34 PM

#

it just responded after long lag

#

Hi! How can I help you today?
butters@mac ! %

rain loomBOT Apr 20, 2026, 11:35 PM

#

Perfect, that’s the answer.

Your Claude login works, so the fix is:

Run:
```
openclaw onboard
```
When it asks about Anthropic auth, choose:
```
Anthropic Claude CLI
```
not setup-token
Finish onboarding

#

Restart OpenClaw:
```
openclaw gateway restart
```
Try Discord again

Why: claude -p working means your login + extra usage are fine.
So OpenClaw should use the Claude CLI auth path, not the old setup-token path.

#

Yep, that’s good, Claude CLI is working. 🎉

Now do this:

openclaw onboard

When prompted, choose:

Anthropic Claude CLI

Then restart:

#

openclaw gateway restart

Then test Discord again.

manic compass Apr 20, 2026, 11:36 PM

#

use existing values?

rain loomBOT Apr 20, 2026, 11:37 PM

#

Yes, use the existing values.

Just change the Anthropic auth choice to:

Anthropic Claude CLI

And do not choose Reset.

manic compass Apr 20, 2026, 11:38 PM

#

Model/auth provider
│ ● Anthropic (API key + legacy token)
│ ○ BytePlus
│ ○ Chutes
│ ○ Cloudflare AI Gateway
│ ○ Copilot Proxy
│ ○ Custom Provider
│ ○ DeepSeek
│ ○ Fireworks
│ ○ Gemini CLI OAuth
│ ○ GitHub Copilot
│ ○ Google
│ ○ Hugging Face
│ ○ Kilo Gateway
│ ○ LiteLLM
│ ○ Microsoft Foundry
│ ○ MiniMax
│ ○ Mistral AI
│ ○ Moonshot AI (Kimi K2.5)
│ ○ Ollama
│ ○ OpenAI
│ ○ OpenAI Codex
│ ○ OpenCode
│ ○ OpenRouter
│ ○ Qianfan
│ ○ Qwen Cloud
│ ○ SGLang
│ ○ StepFun
│ ○ Synthetic
│ ○ Together AI
│ ○ Venice AI
│ ○ Vercel AI Gateway
│ ○ vLLM
│ ○ Volcano Engine
│ ○ xAI (Grok)
│ ○ Xiaomi
│ ○ Z.AI
│ ○ Skip for now

rain loomBOT Apr 20, 2026, 11:38 PM

#

Yep, pick:

Anthropic (API key + legacy token)

Then on the next Anthropic screen, choose the Claude CLI option.

manic compass Apr 20, 2026, 11:39 PM

#

its either api key or set up token

rain loomBOT Apr 20, 2026, 11:40 PM

#

Then don’t pick setup-token again. That’s the path that’s already failing.

Do this instead:

Exit that wizard screen
Run:
```
openclaw configure
```
See if Claude CLI shows up there

If Claude CLI appears, pick that.
If it still only shows API key or setup-token, then your current build isn’t exposing the Claude CLI path there yet.

In that case, your practical options are:

use Anthropic API key now, or

#

stay on setup-token and keep hitting the Anthropic extra-usage issue

So the important part is: don’t re-select setup-token.

manic compass Apr 20, 2026, 11:43 PM

#

Select sections to configure
│ ● Workspace (Set workspace + sessions)
│ ○ Model
│ ○ Web tools
│ ○ Gateway
│ ○ Daemon
│ ○ Channels
│ ○ Plugins
│ ○ Skills
│ ○ Health check
│ ○ Continue
└

rain loomBOT Apr 20, 2026, 11:44 PM

#

Pick:

Model

That’s the section where the auth/provider choice lives.

manic compass Apr 20, 2026, 11:45 PM

#

Anthropic auth method
│ ● Anthropic API key (Direct Anthropic API key)
│ ○ Anthropic setup-token
│ ○ Back

rain loomBOT Apr 20, 2026, 11:46 PM

#

That confirms it, your current build is only offering:

API key
setup-token

So:

If you have an Anthropic API key, choose Anthropic API key
If you do not have one, choose Back and do not pick setup-token again

Because setup-token is the path that’s already failing for you.

So the practical fix right now is an Anthropic API key.
If you specifically want login/Claude CLI auth, you’ll need a build/path that exposes that option.

manic compass Apr 20, 2026, 11:48 PM

#

i just asked claude and it said this. i think im at step 2

Try these in order:

Verify extra usage is re-enabled. Go to claude.ai → Settings → Billing (or the Usage section). After topping up, the "Allow extra usage" toggle sometimes flips off or needs re-confirmation. Make sure it's on and shows your new balance.
Refresh your OAuth token. If OpenClaw authenticates via your Claude Max subscription (rather than a raw API key), the token may have gone stale when requests started failing. In your OpenClaw setup, re-run whatever login/auth command you originally used (often something like claude login or a re-auth flow in the harness config).
Restart the bot process. Kill the OpenClaw process entirely and relaunch it. Cached 429s or backoff timers inside the harness won't clear on their own — some HTTP clients hold onto rate-limit state for a while.
Check the actual error in Discord/logs. The specific error message matters a lot here. If it says something like credit_balance_too_low, rate_limit_exceeded, invalid_api_key, or 401 unauthorized, that narrows it down fast. Can you grab the exact error text from your terminal or Discord channel where OpenClaw is reporting?
If you're using an API key instead of Max auth: Log into console.anthropic.com, confirm the key is still active and the workspace has credit, and that you haven't hit a per-key spend limit.

rain loomBOT Apr 20, 2026, 11:49 PM

#

You’re basically past step 2 already.

Here’s where you actually are:

Step 2 (refresh auth): done enough
We re-did auth, and your Anthropic probe came back ok.
Step 3 (restart): done
You already restarted the gateway.

Step 4 (check real error): done
We got the exact log:

400 invalid_request_error: You're out of extra usage...

So the only useful item left from that list is step 1:

#

Do this next

Go to:

https://claude.ai/settings/usage

and confirm:

Allow extra usage is actually ON
your topped-up balance is visible there

If that is already ON

Then the setup-token path is still being rejected by Anthropic on real model calls.

At that point, your practical options are:

Use an Anthropic API key in OpenClaw now, or
Wait / contact Anthropic, because the login-backed extra-usage lane still isn’t being accepted for your actual requests

#

So no, I would not keep repeating setup-token / restart again. We already proved the token itself is valid enough to probe.

manic compass Apr 20, 2026, 11:50 PM

#

yes extra usage is on

rain loomBOT Apr 20, 2026, 11:50 PM

#

Then we’ve pretty much isolated it.

What we know:

your token probes OK
claude -p works
extra usage is ON
but real OpenClaw Anthropic requests still get:
```
You're out of extra usage
```

So this is not a Discord issue anymore. It looks like a setup-token / Anthropic-side entitlement issue on real model calls.

Your practical options now

1. Fastest fix: use an Anthropic API key

#

In OpenClaw, switch Anthropic auth to API key.

2. If you want to stay login-based

You’ll probably need to wait or contact Anthropic, because their side is still rejecting the actual OpenClaw requests even though your login itself works.

One OpenClaw note

This log classification also looks wrong:

failoverReason: "format"

That’s probably a bug on our side. If you want to report it, include the log lines here:

https://github.com/openclaw/openclaw/issues

So, bluntly: if you need it working now, use an Anthropic API key.

manic compass Apr 20, 2026, 11:50 PM

#

Last login: Mon Apr 20 16:43:11 on ttys000
You have new mail.
butters@mac ~ % claude login
Claude Code v2.1.92
▗ ▗ ▖ ▖ Opus 4.6 (1M context) · Claude Max
/Users/butters
▘▘ ▝▝ [✻] [✻] [✻] · Share Claude Code and earn $10 of extra usage · /passes

↑ Opus now defaults to 1M context · 5x more room, same pricing

❯ login

⏺ What are you trying to log in to? For example:

Claude Code — run ! claude login or /login in the CLI
GitHub — run ! gh auth login
A cloud provider (AWS, GCP, Azure) — let me know which one

Could you clarify what service you need to authenticate with?

────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
❯
───────────────────

rain loomBOT Apr 20, 2026, 11:51 PM

#

Yep, that’s the wrong command path.

You opened Claude Code and then typed login inside it.
If you actually want to refresh Claude CLI auth, the command is:

claude auth login

But honestly, based on everything we already tested, your Claude login is already fine. claude -p "hi" proved that.

So I don’t think re-login is the real fix here.

Bottom line

At this point your real choices are:

Use an Anthropic API key in OpenClaw if you need it working now, or

#

Wait / contact Anthropic about the extra-usage rejection on the setup-token path

So you can Ctrl+C out of that screen.

manic compass Apr 21, 2026, 12:24 AM

#

discor is back working i think but so slow

#

what is the discord command to reboot and restart thes sytem

rain loomBOT Apr 21, 2026, 12:25 AM

#

Nice, that’s progress. 🎉

If it’s working but slow, the fastest check is:

try a brand new Discord thread / fresh message
send something tiny like:
```
hi
```

If the fresh thread is still slow, it’s probably Anthropic-side latency, not Discord.

If the fresh thread is faster, then the slowness is probably from session/context history in the current thread.

So I’d test a fresh thread first before changing anything else.

#LLM request rejected: You're out of extra usage. Add more at claude.ai/settin