You're asking how to handle LLM API call errors when using Azure OpenAI. Below are concise, actionable options you can use with LiveKit Agents to make your agent resilient and user-friendly. 1) Use FallbackAdapter for automatic failover • Chain providers so requests automatically retry on the next provider when Azure fails. • Example (Python):

from livekit.agents import llm
from livekit.plugins import openai

session = AgentSession(
  llm=llm.FallbackAdapter([
    openai.LLM(model="gpt-4o"),                        # primary
    openai.LLM.with_azure(model="gpt-4o", ...),        # fallback
  ]),
)

• Failed providers are marked unhealthy and periodically rechecked. 2) Listen for ErrorEvent and handle errors programmatically • Subscribe to the session

error

event to log, notify users, or trigger recovery. • Example (Python):

@session.on("error")
def on_error(event):
    if not event.error.recoverable:
        # Fatal error (e.g., auth failure)
        print(f"Unrecoverable error: {event.error}")
        session.say("Sorry, I'm having trouble processing your request.")
    else:
        # Transient error (e.g., rate limit)
        print(f"Transient error: {event.error}")

• Useful properties:

error.recoverable

,

source

,

model_config

. 3) Ensure authentication and environment variables are correct • Set Azure keys/endpoints and API version:

AZURE_OPENAI_API_KEY=your-key
AZURE_OPENAI_ENDPOINT=<https://your-endpoint.openai.azure.com>
OPENAI_API_VERSION=2024-10-01-preview

• You can also use a Microsoft Entra ID token if needed. Quick best-practice checklist • Chain providers with

FallbackAdapter

for high availability. • Subscribe to

error

events and distinguish recoverable vs unrecoverable errors. • Inform users gracefully (e.g.,

session.say()

). • Validate API keys, endpoint, and API version early. If you want, I can: • Show a more complete code example for your language (Python or JS). • Help design retry/backoff logic or logging strategies tailored to your app. Sources: Events and error handling | LiveKit docs | Azure OpenAI LLM plugin guide | LiveKit docs