Skip to content

QUOTA_ISSUE_FIX.md

Latest commit

 

History

History
176 lines (126 loc) · 3.97 KB

QUOTA_ISSUE_FIX.md

File metadata and controls

176 lines (126 loc) · 3.97 KB

Fixing the Quota Issue

What Happened

You got this error:

429 RESOURCE_EXHAUSTED
Quota exceeded for metric: gemini-2.0-flash-exp
limit: 0

Why: The experimental model gemini-2.0-flash-exp has a quota limit of 0 for some API keys.

Solution: Use a stable model instead!

✅ Fixed!

I've updated all files to use gemini-1.5-flash (stable, fast, free tier).

🚀 Try Again

Option 1: Edit the test file (Easiest)

  1. Open test_with_your_key.py
  2. Replace the API_KEY at the top with your full key: 
    API_KEY = "AIzaSyAORX...pJwY"  # Replace with YOUR full key
  3. Save and run: 
    python test_with_your_key.py

Option 2: Run directly

python test_llm_simple.py YOUR_FULL_API_KEY

Replace YOUR_FULL_API_KEY with your actual key starting with AIzaSyA...

Option 3: Export and run

export GEMINI_API_KEY='AIzaSyAORX...pJwY'  # Your full key
python test_llm_simple.py

📊 Available Models (Free Tier)

All files now default to gemini-1.5-flash:

Model Speed Quality Free Tier
gemini-1.5-flash ✅ Fast Good 15 RPM, 1M TPD
gemini-1.5-pro Slow Better 2 RPM, 32K TPD
gemini-pro Medium Good 15 RPM, 32K TPD
gemini-2.0-flash-exp Fast Great ❌ 0 (not available)

Legend:

  • RPM = Requests per minute
  • TPD = Tokens per day

🎯 Recommended: Use gemini-1.5-flash

This is now the default in all files:

  • ✅ Fast response (~500ms)
  • ✅ Good quality
  • ✅ Free tier: 15 requests/min, 1M tokens/day
  • ✅ Stable and reliable

🔧 Check Your Quota

Visit: https://aistudio.google.com/app/prompts/new_chat

  1. Try a simple prompt: "Hello"
  2. If it works → Your API key is good!
  3. Check quota at: https://console.cloud.google.com/apis/api/generativelanguage.googleapis.com/quotas

📝 Quick Test

Run this to verify your key works:

python -c "
import os
from google import genai

os.environ['GEMINI_API_KEY'] = 'YOUR_KEY_HERE'
client = genai.Client(api_key=os.environ['GEMINI_API_KEY'])

response = client.models.generate_content(
    model='gemini-1.5-flash',
    contents='Say hello'
)

print('✅ API key works!')
print('Response:', response.text)
"

Replace YOUR_KEY_HERE with your actual key.

🎉 Next Steps

Once you verify the key works:

  1. Test the summarization:

    python test_llm_simple.py YOUR_KEY

  2. Run the coordinator:

    export GEMINI_API_KEY='YOUR_KEY'
python src/agents/adk_enhanced/coordinator.py

  3. Deploy on Pi:

    # Copy to Pi
scp -r src/agents/adk_enhanced pi@raspberrypi:~/object-tracking-agent/src/agents/

# On Pi
export GEMINI_API_KEY='YOUR_KEY'
python src/agents/adk_enhanced/coordinator.py

💡 Model Selection

You can override the model with:

# For testing
python test_llm_simple.py YOUR_KEY

# For coordinator
ADK_MODEL=gemini-1.5-pro python src/agents/adk_enhanced/coordinator.py

Available models:

  • gemini-1.5-flash (default, recommended)
  • gemini-1.5-pro (better quality, slower)
  • gemini-pro (older, still good)

🐛 Still Having Issues?

Try these in order:

  1. Generate a new API key:

  2. Check API is enabled:

  3. Wait and retry:

    • Sometimes takes a few minutes for new keys to activate
    • Wait 5 minutes and try again

  4. Use rule-based:

    • System works fine without LLM!
    • Just won't have natural language summaries
    • All other features work perfectly

✅ Summary

  • ❌ gemini-2.0-flash-exp → Has quota limit of 0
  • ✅ gemini-1.5-flash → Works, fast, free tier
  • ✅ All files updated to use stable model
  • ✅ Ready to test again!

Run: python test_with_your_key.py (easiest)