Chat Fix Status - January 1, 2026

Status: Code fix committed, deployment workflow updated, waiting for deployment

✅ Completed Fixes

1. Code Fix (Committed)

  • File: backend/agents/base.py
  • Changes:
    • Use max_completion_tokens instead of max_tokens for gpt-5.1-chat
    • Remove temperature parameter for gpt-5.1-chat (only default value 1 is supported)
  • Commit: Already in codebase

2. Deployment Workflow Fix (Just Committed)

  • File: .github/workflows/deploy.yml
  • Change: Set azureAiModelRouter= (empty string) instead of azureAiModelRouter=model-router
  • Commit: b40491a18 - “fix: Disable Model Router in deployment workflow”
  • Status: ✅ Committed and pushed

⏳ Next Steps

Step 1: Wait for Active Deployment to Complete

Current Status:

  • There’s an active deployment that caused a conflict
  • The failed deployment (00:01:39Z) was blocked by active deployment (started 00:01:56Z)
  • Wait for the active deployment to complete or fail

Check Status:

gh run list --workflow=deploy.yml --limit 1

Step 2: Wait for CI to Complete

The push will trigger:

  1. CI workflow - Builds and tests the code
  2. Deploy workflow - Deploys to Azure (only runs after CI succeeds)

Wait for: CI workflow to complete successfully

Step 3: Monitor Deployment

Once CI completes and Deploy workflow starts:

gh run watch

Expected:

  • ✅ Infrastructure deployment succeeds (Model Router disabled)
  • ✅ Backend container deploys with the code fix
  • ✅ Configuration: AZURE_AI_MODEL_ROUTER is empty/not set

Step 4: Verify Configuration

After deployment succeeds, verify Azure Container Apps:

az containerapp show \
  --name staging-env-api \
  --resource-group zimax-ai \
  --query "properties.template.containers[0].env[?name=='AZURE_AI_MODEL_ROUTER' || name=='AZURE_AI_DEPLOYMENT' || name=='AZURE_AI_API_VERSION']" \
  --output table

Expected:

  • AZURE_AI_MODEL_ROUTER = (not present or empty) ✅
  • AZURE_AI_DEPLOYMENT = gpt-5.1-chat ✅
  • AZURE_AI_API_VERSION = 2024-12-01-preview ✅

Step 5: Test Chat Endpoint 

python3 scripts/test-chat-debug.py \
  --token "YOUR_JWT_TOKEN" \
  --message "hi"

Expected Result:

  • HTTP 200 OK
  • Actual agent response (not error message)
  • tokens_used > 0 (indicates LLM call succeeded)

📋 Checklist

  • Code fix committed (max_completion_tokens, no temperature for gpt-5.1-chat)
  • Deployment workflow updated (Model Router disabled)
  • Changes pushed to main branch
  • CI workflow completes successfully
  • Active deployment completes (wait if needed)
  • Deploy workflow triggers after CI
  • Infrastructure deployment succeeds
  • Backend container deploys with fix
  • Verify AZURE_AI_MODEL_ROUTER is empty in Azure
  • Test chat endpoint - should work! ✅

🔍 How the Fix Works

Before (Broken)

  1. Model Router enabled → Used wrong endpoint format
  2. max_tokens parameter → gpt-5.1-chat doesn’t support this
  3. temperature parameter → gpt-5.1-chat doesn’t support custom temperature
  4. Result: 400 Bad Request from LLM API

After (Fixed)

  1. Model Router disabled → Direct model deployment (gpt-5.1-chat)
  2. max_completion_tokens parameter → ✅ Supported by gpt-5.1-chat
  3. No temperature parameter → ✅ Uses default (1)
  4. Result: 200 OK with actual response ✅

⚠️ Important Notes

  1. 14-Minute Rule: Don’t trigger multiple deployments within 14 minutes
  2. Wait for Active Deployment: The previous deployment must complete before new one
  3. Configuration Consistency: Ensure Azure Container Apps matches the deployment parameters
  4. Test After Deploy: Always test chat endpoint after deployment succeeds

📚 Related Documentation

  • docs/troubleshooting/gpt-5.1-chat-api-parameters-fix.md - Code fix details
  • docs/troubleshooting/deployment-failure-chat-fix.md - Deployment failure details
  • docs/troubleshooting/chat-deployment-issues-summary.md - Complete summary
  • docs/configuration/config-alignment.md - Configuration reference