OpenRouter "User not found" error that won't go away

[brittle turtle]

Getting AI_APICallError: User not found with OpenRouter across ALL
models (free + paid). API key is valid, I have added credits to my
account, but every model returns this error.

Discovered it's actually HTTP 429 rate limiting, but OpenCode shows
wrong error message. Even paid models are rate limited.

Has anyone else experienced this? Tried waiting, different models,
adding credits, nothing works. Any solutions?

Update: Confirmed via direct API calls - it's definitely rate
limiting, not authentication.

[rare flower]

I'm an AI assistant - my answers may be incorrect and I cannot see screenshots. Please verify information with the OpenRouter docs.

The "User not found" error you're seeing in OpenCode is actually a misreported HTTP 429 (Too Many Requests) rate limiting error. This can happen across all models (both free and paid) and is typically due to exceeding rate limits at the provider or global account level.

Key points:

• The correct error code is 429, which means "Too many requests"—not authentication failure.
• This can occur even with valid credits and API keys.
• Upstream providers (including paid ones) may apply their own rate limits, which can restrict your access even if you have credits.
• "When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it." You may be able to adjust your request filters to allow more routing flexibility. Learn more about load balancing and customization options in the docs: Provider Routing.
• If you continue to get rate-limited, you may need to wait before trying again, lower your request frequency, or contact OpenRouter support for assistance.

For persistent issues or possible misconfiguration, contact the mods through the Discord server for direct support.

Reference: "429 - Too many requests! ... When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it." Llama 3.3 Nemotron Super 49B v1

-# This is an automated response powered by Answer HQ. Use !mute to stop the bot from responding.

[brittle turtle]

Still getting the same error with Llama 3.3:

• Tried openrouter/meta-llama/llama-3.3-70b-instruct:free
• Same AI_APICallError result
• Logs show model is found and initialized correctly
• API call is made but fails with same rate limiting issue

The specific error pattern:
INFO: providerID=openrouter
modelID=meta-llama/llama-3.3-70b-instruct:free found
ERROR: AI_APICallError with
url=https://openrouter.ai/api/v1/chat/completions

This happens across ALL models I've tested (free AND paid), even
after adding credits. The issue appears to be systematic rate
limiting rather than model-specific routing problems.

[rare flower]

You're correct—this is a systematic rate limiting issue, not a problem with your account, credits, or specific model. The error is an HTTP 429 "Too many requests" response, which can occur across all models, including both free and paid options, if rate limits are exceeded at the provider or global level.

According to the documentation:
"429 - Too many requests! ... When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it." You might be able to resolve some rate limit errors by adjusting your request filters to allow routing to more providers. Learn more about this in the Provider Routing documentation.

If you're still rate limited on all models after waiting, lowering frequency, or adjusting filters, this suggests either:

• A global or account-wide rate limit is being enforced, or
• All available providers for your requests are rate-limited simultaneously.

Since this is affecting all models and persists after adding credits, you should contact mods through the Discord server for further assistance.

Source:
"429 - Too many requests! ... When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it." Llama 3.3 Nemotron 49B v1

-# This is an automated response powered by Answer HQ. Use !mute to stop the bot from responding.