.sundancekid.
01/25/2026, 11:38 AMAI Evy
01/25/2026, 11:38 AMAI Evy
01/25/2026, 11:39 AMAI Evy
01/25/2026, 11:39 AMsip:sip.retellai.com [(1) ]()
## Current SIP Configuration Details
The documented SIP infrastructure includes:
- **SIP server URI**: sip:sip.retellai.com
- **IP block for traffic**: 18.98.16.120/30 (All regions) , 143.223.88.0/21 (certain United States traffic) , 161.115.160.0/19 (certain United States traffic)
**Supported transports**:
- TCP (Recommended)
- UDP
- TLS
For different transports, you can postpend the transport method:
- TCP: sip:sip.retellai.com;transport=tcp
- UDP: sip:sip.retellai.com;transport=udp
- TLS: sip:sip.retellai.com;transport=tls
## Regarding Region-Specific SIP Endpoints
I don't have information in the available sources about region-specific SIP URIs (such as ie.sip.retellai.com or similar European/African endpoints) . The documentation only references the single global SIP server address.
Your observation that calls fail when using IE1 routing aligns with the fact that the documented IP blocks appear to be US-based.
## Recommendations
Since the sources don't document regional SIP endpoints or plans for them, I'd suggest:
1. Contact Retell AI directly at support@retellai.com to inquire about regional SIP ingress availability or roadmap
2. Check out Retell AI's telephony partners like [Jambonz]() or [Cloudonix]() , which can manipulate SIP call flow and may offer regional routing options
Would you like a general answer about potential latency optimization strategies, or would you prefer to reach out to Retell AI support for specifics on regional infrastructure?AI Evy
01/25/2026, 11:39 AMAI Evy
01/25/2026, 11:42 AM.sundancekid.
01/25/2026, 11:42 AMAI Evy
01/25/2026, 11:42 AMAI Evy
01/25/2026, 11:43 AMsip:sip.retellai.com [(1) ]()
## What the Documentation Shows for Latency Optimization
Retell AI's architecture achieves low latency through: [(2) ]()
- Real-time speech recognition that processes audio streams as they arrive
- Optimized LLM routing with direct connections to language models
- Edge computing with distributed processing
- Predictive response generation that begins formulating responses before the user finishes speaking
Production voice AI agents typically aim for 800ms or lower latency to maintain conversational flow. [(3) ]()
## Suggestions Given Current Constraints
Since regional SIP endpoints are not documented:
1. Use a localized termination URI near your region when configuring Twilio—the documentation mentions this as a troubleshooting step for outbound issues. [(4) ]()
2. Consider Retell AI's telephony partners such as [Jambonz]() or [Cloudonix]() , which can manipulate SIP call flow and may offer regional routing options. [(1) ]()
3. Verify your Twilio configuration doesn't have any spaces in the Termination SIP URI. [(4) ]()
4. **ContacAI Evy
01/25/2026, 11:43 AMAI Evy
01/25/2026, 11:43 AMAI Evy
01/25/2026, 11:43 AM