[Bug]: GENERATION_FAILED (fp: 7a92d5b3)

[dsoasti]

Description

I'd love to write something but I have no idea why this window opened

Steps to reproduce

I'd love to write something but I have no idea why this window opened

Expected behavior

I'd love to write something but I have no idea why this window opened

Actual behavior

Error invoking remote method 'codesign:v1:generate': CodesignError: 429 Provider returned error
meta-llama/llama-3.3-70b-instruct:free is temporarily rate-limited upstream. Please retry shortly, or add your own key to accumulate your rate limits: https://openrouter.ai/settings/integrations

open-codesign version

0.2.0

Platform

Windows

OS version

Windows 11 (10.0.26200)

Model provider (if relevant)

Other

Error code

GENERATION_FAILED

Diagnostics bundle

Bundle saved locally at ~\Downloads\open-codesign-diagnostics-2026-07-08T19-08-01.zip. Attach it to this issue after submitting.

Logs

}
}
[2026-06-19 18:08:59.499] [info]   (main:ipc) [generate] step=send_request {
  runId: 'mqljfp46-gvjwhd',
  generationId: 'mqljfp46-gvjwhd',
  provider: 'custom-openrouter-z1zq',
  modelId: 'meta-llama/llama-3.3-70b-instruct:free'
}
[2026-06-19 18:08:59.503] [info]   (main:ipc) agent.turn_start { runId: 'mqljfp46-gvjwhd', generationId: 'mqljfp46-gvjwhd' }
[2026-06-19 18:10:00.518] [info]   (main:ipc) agent.turn_end {
  runId: 'mqljfp46-gvjwhd',
  generationId: 'mqljfp46-gvjwhd',
  deltas: 0,
  tools: 0
}
[2026-06-19 18:10:00.533] [info]   (main:ipc) agent.end { runId: 'mqljfp46-gvjwhd', generationId: 'mqljfp46-gvjwhd' }
[2026-06-19 18:10:00.535] [error]  (main:ipc) [generate] step=send_request.fail {
  runId: 'mqljfp46-gvjwhd',
  generationId: 'mqljfp46-gvjwhd',
  provider: 'custom-openrouter-z1zq',
  modelId: 'meta-llama/llama-3.3-70b-instruct:free',
  ms: 61034,
  stopReason: 'error'
}
[2026-06-19 18:10:00.540] [error]  (main:ipc) generate.fail {
  runId: 'mqljfp46-gvjwhd',
  generationId: 'mqljfp46-gvjwhd',
  ms: 214814,
  provider: 'custom-openrouter-z1zq',
  modelId: 'meta-llama/llama-3.3-70b-instruct:free',
  baseUrl: 'https://openrouter.ai/api/v1',
  status: 429,
  message: '429 Provider returned error\n' +
    'meta-llama/llama-3.3-70b-instruct:free is temporarily rate-limited upstream. Please retry shortly, or add your own key to accumulate your rate limits: https://openrouter.ai/settings/integrations',
  code: 'PROVIDER_ERROR'
}
[2026-06-19 18:10:00.832] [error]  (renderer:console) [store] applyGenerateError {"generationId":"mqljfp46-gvjwhd","designId":"e58314c3-7a5b-4f47-8ead-da1214d70dbb","message":"Error invoking remote method 'codesign:v1:generate': CodesignError: 429 Provider returned error\nmeta-llama/llama-3.3-70b-instruct:free is temporarily rate-limited upstream. Please retry shortly, or add your own key to accumulate your rate limits: https://openrouter.ai/settings/integrations"} {}
[2026-07-08 11:07:55.032] [info]   [boot] open-codesign starting {
  version: '0.2.0',
  platform: 'win32',
  electron: '39.8.9',
  node: '22.22.1'
}
[2026-07-08 11:07:58.151] [info]   (main:boot) templates.ensure {
  action: 'skipped',
  source: 'C:\\Users\\David Soasti\\AppData\\Local\\Programs\\Open CoDesign\\resources\\templates',
  dest: 'C:\\Users\\David Soasti\\AppData\\Roaming\\@open-codesign\\desktop\\templates',
  copiedFiles: 0,
  updatedFiles: 0
}

[github-actions]

Thank you for the report, @dsoasti. This is not a bug in Open CoDesign itself — the error you're seeing is a HTTP 429 (Too Many Requests) from OpenRouter for the free-tier model meta-llama/llama-3.3-70b-instruct:free.

OpenRouter's free models have rate limits per IP/key. The error message suggests either waiting a bit and retrying, or adding your own API key to avoid free-tier throttling.

What you can do:

1. Wait and retry – rate limits usually reset within minutes, but free tiers can be congested.
2. Add your own OpenRouter API key – go to your provider settings in Open CoDesign (gear icon → "API Configuration" or similar) and enter a key from OpenRouter's settings page. This gives you dedicated rate limits.
3. Switch to a different model – you can choose a non-free model or a different provider altogether (e.g., Claude, GPT, Gemini) using your own credentials.

If you need help finding the provider configuration, the quickstart docs cover provider setup. The app stores config locally in ~/.config/open-codesign/config.toml.

Let us know if you keep seeing this after adding a key, or if you run into a different error.

---

Open-CoDesign Bot

[github-actions]

Thanks for the follow-up suggestion, @tyler-ross681. You're right that a built-in retry and fallback mechanism would be a helpful improvement for handling rate-limited providers like OpenRouter's free tier.

Currently, Open CoDesign does not implement automatic retry or provider failover on HTTP 429 errors. The error is passed through directly from the pi-ai layer to the user, and the generation fails. Users can manually:

• Wait and retry (rate limits usually reset within minutes)
• Switch to a different model or provider in the settings
• Add their own API key to avoid free-tier throttling

There is no "AI Badgr gateway" or similar retry/routing middleware in the codebase as of v0.2.0. However, this is a valuable feature request — you're welcome to open a feature request or discuss it in Discussions to gather community interest.

For now, the workarounds mentioned in the earlier response remain the best path for users hitting this error.

---

Open-CoDesign Bot