Skip to main content

Troubleshooting

Diagnostic Logging​

By default, Orion only prints warnings and errors to the Output Log to keep it clean. When troubleshooting, you can enable verbose logging to see detailed diagnostic output.

Console Commands (type in the Output Log command bar or press ~ to open the console):

CommandWhat it does
Azimuth.Orion.EnableLoggingEnables verbose logging — shows all Orion diagnostic output
Azimuth.Orion.DisableLoggingRestores default logging (warnings and errors only)
Azimuth.Orion.StatusPrints current plugin configuration: version, selected model, enabled providers, feature toggles

When to use:

  • If you receive API errors and need to see the full request/response flow
  • If a feature isn't behaving as expected and you need to see internal state
  • When submitting a support ticket — enable verbose logging, reproduce the issue, then share the Output Log

All commands auto-complete in the UE5 console.

"API request failed (Code: 401)"​

Cause: Invalid or expired API key

Solution:

  1. Open Settings > AI Providers > [Your Provider]
  2. Check that the API key is entered correctly
  3. Click Validate button
  4. If validation fails:
    • Go to the provider's website (console.x.ai, platform.openai.com, etc.)
    • Check that your API key is still active
    • Check that you have available credits/billing set up
    • Generate a new API key if needed
  5. Enter the new key in Orion, validate again

"API request failed (Code: 429)"​

Cause: Rate limit exceeded (too many requests too fast)

Solution:

  • Wait 30-60 seconds, then try again
  • Orion has client-side rate limiting, but providers have their own limits
  • Check your provider dashboard for rate limit info
  • Consider upgrading to a higher tier plan if you hit limits frequently

"Conversation chain expired on the server"​

Cause: The xAI conversation chain expired (typically after extended periods of inactivity)

Solution:

  • This is handled automatically—Orion clears the stale conversation and starts fresh
  • Just resend your message
  • To prevent: Use "Clear server history" button in settings after long breaks

"The currently selected model does not support images"​

Cause: You captured a screenshot but your current model doesn't support vision

Solution:

  1. Open Settings > AI Configuration > Model ID
  2. Select a vision-capable model (marked [Vision] in the dropdown), for example:
    • xAI: grok-4, grok-4-1-fast-reasoning, grok-4-1-fast-non-reasoning
    • OpenAI: gpt-5.2, gpt-5-mini, gpt-5.3-codex, o4-mini
    • Anthropic: All Claude models (Opus 4.6, Opus 4.5, Sonnet 4.6, Sonnet 4.5, Haiku 4.5)
    • Google: All Gemini models (3 Pro, 3 Flash, 3.1 Pro, 2.5 Pro, 2.5 Flash)
  3. Capture screenshot again
  4. Send your query

Right-click menu doesn't appear​

Cause: "Enable Right-Click Menu" disabled in settings

Solution:

  1. Settings > Interface
  2. Check Enable Right-Click Menu (turn it ON)
  3. Restart editor (may be required)
  4. Right-click menu should now appear

Response is incomplete or cut off​

Cause: Model hit max output token limit mid-response

Solution:

Option 1: Use Retry button

  1. Click Retry in the chat bottom bar
  2. AI may continue from where it stopped (works with some models)

Option 2: Increase token limit

  1. Settings > AI Configuration > Max Output Tokens
  2. Increase to 8000 or 16000 (or 0 for unlimited)
  3. Ask your question again

Option 3: Ask more focused questions

  • Instead of "Explain everything about Blueprints", ask "Explain Blueprint execution flow"
  • Narrow the scope to get complete answers

Chat window won't scroll to bottom​

Cause: Manual scroll position is preserved by design (so you can read old messages while new ones arrive)

Solution:

  • Scroll to bottom manually — Auto-scroll re-enables when you're at the bottom
  • Send a new message — Auto-scroll re-enables on send
  • This is intentional behavior (not a bug)

"Failed to compile Blueprint"​

Cause: Blueprint has errors, and "Compile On Send" is enabled

Solution:

  • This is actually helpful—the AI receives the compiler errors
  • Ask: "Why did compilation fail?" or "What are these errors?"
  • AI will diagnose based on the error messages
  • Fix the errors, send again

If you don't want auto-compilation:

  • Settings > Blueprint Context > Compile On Send (uncheck)

"Request timeout"​

Cause: The AI provider took longer than the timeout setting to respond

Solution:

  1. Increase timeout: Settings > AI Configuration > Request Timeout (Seconds)
  2. Set to 90 or 120 seconds (default is 60-75 depending on model)
  3. Try again
  4. If timeouts persist:
    • Switch to a faster model (grok-code-fast-1, gpt-5-mini)
    • Reduce Blueprint/context size (send Selection instead of Full Graph)
    • Check your internet connection