Inventory the trunk, numbers and Origination URLs separately
Twilio routes origination calls from associated telephone numbers toward the Origination SIP URLs owned by one Trunk SID. Record the number association, Trunk SID and every OriginationUrl SID, friendly name, sip: URI, enabled state, priority, weight and update time. A number existing in the account does not prove it is associated with the intended trunk, and a created URL does not prove it is enabled or selected.
Elastic SIP Trunking origination is incoming traffic from the PSTN toward the customer PBX or SBC. Keep it distinct from termination, where the customer sends outbound calls to Twilio, and from the separate Twilio SIP Interface registration model.
Documented service facts
Reviewed 2026-10-17. Compare SIP trunk services →
Outbound authentication
DocumentedTermination requires the personal Termination SIP URI plus at least an IP Access Control List or Credential List. When both are configured, both are enforced; an attached Credential List produces a 407 challenge before the authenticated retry.
SIP registration
DocumentedTwilio directs the communications infrastructure not to register for this trunk. Credentialed termination and optional SIP OPTIONS health checks remain distinct from endpoint registration.
Inbound delivery
DocumentedAssociate numbers with the trunk and configure enabled OriginationUrl resources with explicit `sip:` URI, priority and weight. The current API page presents conflicting default markers, so preserve stored values rather than relying on defaults.
Inbound failover
DocumentedLower numeric priority is more important and weight distributes only among same-priority URLs. Current documentation lists response classes that stop advancement and a four-second no-response boundary; a Disaster Recovery URL is a later TwiML path after the SIP URLs do not succeed.
Number and regional availability
Not verifiedInternational number selection is documented; exact country, number inventory and port eligibility for your account are not verified.
Audio codecs
DocumentedPCMU and PCMA are generally available; G.729, Opus and AMR-NB have limited availability.
T.38 fax
Not verifiedNot verified for this exact service in this review. Do not infer fax support from another Twilio product.
Signaling and media encryption
DocumentedThe per-trunk Secure Trunking control documents TLS for signaling and SRTP for media. Certificate trust, negotiated protection and every PBX-facing leg still require verification.
Call-start capacity
DocumentedDefault termination is 1 new call per second per trunk per region; Console supports self-service settings up to 5 CPS, with higher or pooled limits handled separately. Published unlimited origination CPS and concurrent-call language is carrier- and Business-Profile-dependent.
Emergency calling
DocumentedDocumented for the US, Canada, UK, Australia, Ireland, France, Germany and Austria. Country-matched caller identity, registered address and callback origination route remain required; 933 is the documented automated test path for the US and Canada.
Pricing model
DocumentedUsage-based calling, separate number rental and optional features; destination, country and committed-use terms affect cost.
Administration and support
Not verifiedConsole and REST resources expose Trunks and Origination URLs. Preserve number association, URL SID, enabled state, explicit priority and weight, provider attempts, disaster-recovery webhook and rollback evidence; contracted support response times remain unverified.
SMS and MMS
Not verifiedNumber-specific messaging eligibility, registration requirements and charges are not verified in this trunk review.
Treat priority and weight as different routing controls
The lowest numeric priority is the most important. Weight divides requests only among enabled URLs with the same priority; a larger weight receives a larger relative share. Weight is therefore not a standby sequence, and priority is not proof that every request reached or was accepted by the first destination.
The current OriginationUrl API page says the descriptive defaults for priority and weight are 10 while its generated property markers display Default: 0. Configure both values explicitly, read the returned resource and preserve the effective values rather than relying on either default presentation.
Follow the provider attempt and exact SIP result
Twilio documents response-dependent advancement. Its current configuration page says 2xx, 400, 404, 405, 410, 416, 482, 484, 486 and 6xx responses stop failover to the next Origination SIP URI; no SIP response advances after four seconds. Preserve the target URI, selected edge, source and destination addresses, transport, Call-ID, elapsed time and complete response sequence for each attempt.
Do not generalize that published stop list into a complete promise for every transport error, provisional response, timer, retry or restored path. A 486 is an observed final result, not evidence that a later URI should have been tried.
Use disaster recovery only after the SIP destinations are exhausted
Twilio documents a Disaster Recovery URL that is invoked for each new origination request after none of the configured Origination SIP URIs succeeds. The URL must return TwiML and can play a message, collect voicemail, redirect the call or implement other Programmable Voice behavior. Normal Programmable Voice rates apply to that path.
The webhook request, HTTP result, parsed TwiML, subsequent call leg and caller outcome are separate evidence. A configured URL is not proof it was invoked or that the application returned valid instructions, and this new-call workflow does not establish survival of an already connected SIP call.
Keep edge selection and firewall evidence visible
Twilio says origination requests can egress from different edge locations and advises allowing all published signaling and media addresses and ports, because traffic may use another edge during failure. Record the optional edge parameter, current published ranges, observed packet source, firewall decision, SIP destination and RTP endpoints.
A firewall rule that permits one nearby edge is not a complete failover design. Likewise, allowing all documented ranges does not prove number association, URL selection, PBX matching, codec negotiation or two-way decoded audio.
Validate new-call routing without inventing availability
Test the normal destination, same-priority distribution, lower-priority fallback and disaster-recovery application as separate authorized scenarios. Record each new request, selection, response, PBX match, answer, two-way media, release and return to the normal path. Preserve temporary changes and rollback evidence.
Do not infer a service-level result, exact distribution from a tiny sample, active-call survival, price, account capacity or universal compatibility. No Twilio number, trunk, OriginationUrl, PBX, edge, failover event, webhook, media stream or live call was inspected or tested for this guide.
Trace Twilio origination selection into recovery
Use the Twilio origination diagnostic to prove number association, enabled URLs, explicit priority and weight, each SIP result and any disaster-recovery application.
Sources & applicability
Primary references for the technical details above. Operational examples and planning checklists are VoIP.info editorial guidance.
- Twilio Elastic SIP Trunking configuration ↗
- Twilio OriginationUrl resource ↗
- Twilio Programmable Voice failover best practices ↗
- Twilio Elastic SIP Trunking IP addresses ↗
- Twilio Elastic SIP Trunking troubleshooting ↗
- Twilio Trunk resource ↗
- Twilio Interconnect trunk configuration ↗
- Twilio codec availability ↗
- Twilio scale and account limits ↗
- Twilio termination calls per second ↗
- Twilio emergency configuration ↗
- Twilio US pricing context ↗
- Twilio sending-SIP best practices ↗
- Twilio CredentialList resource ↗
- Twilio error 32201: source IP address not in ACL ↗
Examples require adaptation to your topology. No live PBX or hardware testing is claimed.