Model/auth failures: rate limit, billing, or 'all models failed'

Symptoms

• Messages fail with “all models failed”.
• Provider errors like 401/403 (auth), 429 (rate limit), or billing/credits issues.
• Logs show “No API key found for provider …”.

Cause

Usually one of:

• Missing/expired credentials for the selected provider (auth profiles are per agent).
• You are using a model ref that OpenClaw can’t resolve (wrong provider prefix, wrong model id).
• The provider is rate-limiting or billing is blocked.
• Your configured primary model is down and you have no fallbacks.

Fix

1) Get a fast view of model + auth state

openclaw models status

If you’re using multiple agents, inspect the specific agent:

openclaw models status --agent <agentId>

2) Run live probes (real requests)

openclaw models status --probe

Probes may consume tokens and can trigger rate limits, but they quickly tell you if auth works.

3) Fix missing auth for a provider

Common options:

openclaw models auth login --provider <id>
openclaw models auth paste-token --provider <id>
openclaw models auth setup-token --provider anthropic

Then re-check:

openclaw models status

4) Verify model refs (provider prefix matters)

Model refs are parsed by splitting on the first /. If the model id itself contains / (OpenRouter-style), you must include the provider prefix.

Example:

• Correct: openrouter/moonshotai/kimi-k2
• Wrong (ambiguous): moonshotai/kimi-k2

Set a known-good model:

openclaw models set openrouter/anthropic/claude-sonnet-4-5
openclaw models status

5) Add fallbacks (recommended)

openclaw models fallbacks add <provider/model>
openclaw models fallbacks list

Verify

• openclaw models status --probe shows your default/fallback models are reachable.
• Sending a test message succeeds and the gateway logs show a successful model call.

Related

• If /model says “model not allowed”, that’s an allowlist problem: see the dedicated solution page.