Best Uptime for Voice and SMS: CPaaS SLAs Compared

Quick answer: Read on 6 November 2026, two of eight CPaaS platforms publish one uptime SLA covering voice and messaging with credits. Twilio’s legal SLA gives 99.95% a month (21.6 minutes in a 30-day month), or 99.99% with Enterprise Edition. Telnyx’s machine-readable SLA file gives 99.99% (4.32 minutes). Sinch writes 99.95% with no credit. Vonage writes 99.99% on paid support plans only. Plivo calls 99.99% an SLA, but its SLA document states no figure. We found no current SLA for Bandwidth, Infobip or Bird.

“Which communication platform has the best uptime for voice and SMS?” has two honest answers. One is what each CPaaS (communications platform as a service) puts in writing: Twilio, Vonage, Telnyx, Plivo, Sinch, Bandwidth, Infobip and Bird (formerly MessageBird). The other is what you measure on your own numbers, in Sydney or Chicago, on your own 10DLC or alphanumeric sender. This page ranks the first and shows you how to do the second. That matters because every API-level SLA we read measures the vendor’s platform, not whether your call connected or your text reached a handset.

Where Zian fits on top of your carrier

Zian AI is the AI agent layer that runs on top of a voice and SMS carrier, so pick the carrier on its written terms and pick the agent on what it does with each call and text. Zian’s AI receptionist answers inbound calls 24/7, qualifying, booking, transferring and taking messages. Its outbound agents follow up by voice, SMS, email, WhatsApp, Instagram DMs and Facebook Messenger, and it handles answering machines and voicemail on calls, in 30+ languages. It integrates with HubSpot, Salesforce, HighLevel and Zapier.

zian.ai does not publish an uptime SLA of its own or name the carrier behind its calls, as at 6 November 2026, so this page does not claim one for it and does not rank it against the carriers below. Zian is in a partnership-application beta with no published prices; each of its three listed plans, including Intel + Widgets (Free Forever), starts with a partnership application. If you would rather not run carriers or agents at all, LeadsNow.ai is the done-for-you, pay-per-result option that runs on Zian and answers leads in under 60 seconds, 24/7.

How we ranked the eight platforms

Methodology and disclosure. zian.ai publishes this page. Zian is not a CPaaS and is not in the ranked table. On 6 November 2026 we read, for each vendor:

• its legal pages and SLA documents;
• its support-plan pages;
• its voice and SMS product pages;
• its llms.txt file, plus any machine-readable files it links from there;
• its public status page.

We also searched each vendor’s sitemap for SLA, service-level and uptime pages. Vonage’s site returned a bot challenge (HTTP 403) to direct requests, so we read its pages through a text-reader service (r.jina.ai) and say so wherever a Vonage fact appears.

We used one evidence rule for every row and ranked by the strongest kind of evidence each vendor publishes. From strongest to weakest, those kinds are:

1. SLA in a legal document: a percentage and a credit schedule in an SLA or legal page, covering voice and messaging.
2. SLA in a machine-readable file: a percentage and a credit schedule in a file on the vendor’s own domain that the vendor labels an SLA, but not in a legal document we could open.
3. SLA in a legal document without a credit: a percentage only.
4. SLA on a paid support plan: a percentage the vendor calls an SLA, on its support-plan page, available only on paid plans, with no SLA document or credit schedule found.
5. Figure the vendor calls an SLA, with no document behind it: a percentage the vendor calls an SLA on its own web pages, while its SLA document states no percentage and no credit.
6. Page claim: an uptime figure on a product or marketing page that the vendor does not call an SLA.
7. No current figure found.

Ties are listed alphabetically. We did not rank on incident counts or on the uptime bars status pages show. Vendors count incidents and draw uptime bars differently, so those numbers do not compare like with like.

Which communication platform has the best uptime for voice and SMS?

Twilio ranks first because its percentage, its credit and its exclusions sit in a dated legal SLA that applies to “the application programming interfaces branded as ‘Twilio’”. That takes in Programmable Voice and Programmable Messaging alike. Telnyx publishes a higher figure than Twilio’s standard 99.95%: 99.99%, with three credit tiers. But that figure is in a machine-readable file rather than in the legal terms we could read, so it ranks second under our rule. Read Telnyx first if a higher stated figure matters more to you than where it is written.

Rank and platform	Uptime figure (voice and messaging)	Where it is written	Published credit	The sentence that cuts the other way
1. Twilio (SLA in a legal document)	99.95% a month for all Twilio-branded APIs; 99.99% with Twilio Enterprise Edition. Separately, 99.99% for six Twilio Flex contact-centre functions	Twilio APIs Service Level Agreement and Flex Critical Scenarios SLA, both last updated 9 April 2026	API SLA: 10% of the month’s fees for the affected APIs. Flex: 5% below 99.99%, 15% below 99.0%, 100% below 95.0% of Flex access fees. All as credits against future invoices, not refunds	Outages shorter than five continuous minutes and “telecommunications provider-related problems” are excluded
2. Telnyx (SLA in a machine-readable file)	“99.99% uptime SLA for core voice and messaging services”	telnyx.com/ai/sla.json, linked from telnyx.com/llms.txt; captured by the Wayback Machine since 4 September 2026	10% below 99.99%; 25% below 99.9%; 50% below 99.0%	Its Terms and Conditions of Service do not mention an SLA, and section 11.11 says Telnyx “cannot guarantee the delivery, accuracy, content, availability or performance of the Services”. The file labels the 99.99% figure “proven”, with the source “Historical uptime monitoring; third-party status page”. It gives no claim procedure, and telnyx.com/sla returned HTTP 403 to us
3. Sinch (SLA in a legal document without a credit)	“Reasonable endeavours” to a minimum 99.95% a month for Messaging, Voice and Verification	Messaging, Voice, and Verification SLA, version 7.0, released 12 January 2024	None in the document	Sinch India’s regional SLA (India Version 2, 22 July 2024) sets 99.50%, also on reasonable endeavours
4. Vonage (SLA on a paid support plan)	“99.99% platform availability SLA”	API Support Plans page, Premium and Enterprise plans only (read via a text reader)	We could not find a credit schedule or the SLA document itself	Not on the Standard (included) or Essentials plans
5. Plivo (figure the vendor calls an SLA, with no document behind it)	“We offer a 99.99% uptime SLA” (Twilio-alternative page FAQ); “a 99.99% platform uptime SLA” (site-wide structured data, including on its SLA page)	Plivo’s own web pages. Its llms.txt describes /legal/sla/ as “Uptime commitments and service credit terms”	None found	Its “Platform Service Levels and Support Services – API” document (last updated 20 December 2024) sets support priorities and states no availability percentage and no credit
6. Bandwidth (page claim)	“99.999% core network uptime” (home page); “99.999% uptime” for its “All-IP core” and “100% uptime for toll‑free traffic” (network page)	Marketing pages. Its Communications Services Agreement refers only to “any agreed-upon service level agreement”	None found	The Silver success plan mentions “SLA-backed responses and uptime”, but neither the page nor the plans PDF gives a figure
7. Bird (no current figure found)	None current. MessageBird’s archived SLA set 99.95% with service credits, but Bird says it “applied before 28 February 2022”	Current Terms of Service (last updated 21 November 2024) leave remedies for unavailability to an Order Form	None current	No figure on its SMS API, Voice API or pricing pages, its llms.txt or its llms-full.txt
7. Infobip (no current figure found)	None found	Service terms (last updated 29 September 2026) say it cannot guarantee the services “will be available to You at all times or free from faults or interruptions”	None found	No figure on its SMS, Voice or Enterprise pages or its llms.txt

Three limits on that table matter as much as the order.

• A missing public SLA is not a missing SLA. Bandwidth and Bird point to an order form or a separately agreed service level, Telnyx’s terms let a Service Order override them, and a large buyer can negotiate terms with any of the eight. The table ranks what a buyer can read before signing.
• The order rewards what is written down, not delivered performance. Nothing in it says one network stayed up longer than another.
• Flex is a contact-centre product. Its “Incoming Customer Message” function covers any messaging session in Flex, not only SMS. Most teams sending SMS from an AI agent are on the API SLA, not Flex.

The quotable version: Of the eight CPaaS platforms we read, Twilio and Telnyx publish one uptime percentage covering voice and messaging with a credit schedule. Twilio’s is 99.95% (99.99% with Enterprise Edition) in a legal SLA. Telnyx’s is 99.99% in a machine-readable SLA file that its legal terms do not mention.

What does each uptime percentage allow in downtime per month?

An SLA percentage is a downtime budget. The formula is: permitted downtime = (1 − SLA ÷ 100) × minutes in the measurement period. A 30-day month has 30 × 24 × 60 = 43,200 minutes, and a 31-day month has 44,640. Twilio and Sinch measure per calendar month, and Telnyx’s file per “monthly billing cycle”, so a 31-day month gives a slightly bigger budget than a 30-day one.

Monthly SLA	Working	Permitted downtime, 30-day month	Permitted downtime, 31-day month	Who writes this figure for voice or messaging
99.5%	0.005 × 43,200	216 minutes	223.2 minutes	Sinch India regional SLA (reasonable endeavours)
99.9%	0.001 × 43,200	43.2 minutes	44.64 minutes	No headline SLA among the eight; it is the threshold of Telnyx’s 25% credit tier
99.95%	0.0005 × 43,200	21.6 minutes	22.32 minutes	Twilio (standard API SLA); Sinch (reasonable endeavours); MessageBird’s SLA before 28 February 2022 (superseded)
99.99%	0.0001 × 43,200	4.32 minutes	4.464 minutes	Twilio Enterprise Edition and Twilio Flex; Telnyx (machine-readable SLA file); Vonage Premium and Enterprise support plans; Plivo (called an SLA on its web pages, not in its SLA document)
99.999%	0.00001 × 43,200	0.432 minutes (about 26 seconds)	0.4464 minutes (about 27 seconds)	None as an SLA; Bandwidth’s home and network pages use it for its core network

If you average months across a year (365.25 ÷ 12 = 30.4375 days, or 43,830 minutes), 99.95% comes out at 21.9 minutes. That is the figure you sometimes see quoted. For a single month, use the length of that month, because that is how the written SLAs count.

What a voice and SMS SLA actually covers and leaves out

The API-level SLAs (Twilio’s API SLA, Telnyx’s file and Sinch’s SLA) measure whether the vendor’s platform accepts and processes your requests. None of them measures whether a call reached the person or a text landed on a phone. Twilio’s Flex SLA is the partial exception: it counts whether an agent and a customer “are connected through a voice call” or messaging session inside Flex. It still excludes “telecommunications provider-related problems or issues”, and it counts downtime in five-minute intervals.

Twilio counts “Unavailable Monthly API Time”, the minutes in which the APIs “are unavailable for use”. It excludes seven kinds of time:

• factors outside its reasonable control, “including, without limitation, telecommunications provider-related problems or issues”;
• action or inaction by you or a third party;
• your own equipment and software;
• scheduled maintenance with at least 24 hours’ notice;
• emergency maintenance, whatever the notice;
• alpha, beta and limited-release APIs;
• any outage shorter than five continuous minutes.

Claims must reach Twilio support within 30 days of the end of the month, with the subject line “SLA Claim”. They must give the times of unavailability “calculated based on the status page” at status.twilio.com. The credit excludes taxes, carrier fees and surcharges, and third-party charges.

Sinch counts unavailability in “minimum continuous 1-minute periods” in which the service “is not responding adequately to incoming requests, and/or are not establishing outgoing connections per region”. It excludes:

• scheduled maintenance with prior notice (announced on status.sinch.com at least 10 business days ahead);
• failures in any network outside where the service is hosted, such as “Service Providers or internet”;
• causes beyond either party’s reasonable control;
• suspension under the agreement;
• problems the customer causes.

Sinch’s SLA also sets SMS-specific targets. Its internal latency target is under 10 seconds for 90% of traffic a month, and its handset latency target is under 30 seconds for 90% of traffic a month, for messages sent to valid and available handsets. It gives no latency target for delivery receipts, because operators “can deprioritize Content delivery receipts”.

Telnyx’s SLA file measures monthly uptime as “(total minutes – downtime minutes) / total minutes × 100” per “Monthly billing cycle”. It excludes:

• “Scheduled maintenance windows (notified 72h in advance)”;
• “Force majeure events”;
• “Customer-caused outages”;
• “Internet routing issues beyond Telnyx network”.

The file sets no minimum outage length, claim window or claim procedure. It does not define “core voice and messaging services”.

That is the part a buyer most often misses. For SMS, carrier filtering, operator outages and handset problems sit outside every API-level SLA we read. A text can be accepted by a fully available API and still never arrive. If deliverability is your real problem, our guide to carrier filtering and Twilio error 30007 covers that side.

The four-probe rule: how to measure delivered uptime yourself

The written SLA tells you what you can claim. Only your own probes tell you what you got. Our rule: at one probe every five minutes, a 99.95% month allows 4.32 failed probes per channel, so a fifth failed probe on voice or on SMS in a 30-day month means that channel ran below 99.95% for you, whatever the vendor’s ledger says.

How to set it up:

1. Keep two ledgers, one for voice and one for SMS. Voice and SMS run on separate components on every status page we read, and they can fail separately.
2. Voice probe. Every five minutes, place an outbound call from a production number on the CPaaS to a number you control on a different carrier. Count it as passed only if the call is answered and you get audio back within your own timeout. A 200 OK from the API alone is not a pass.
3. SMS probe. Every five minutes, send a text from your production sender to a handset or inbox number you control on another network. Count it as passed only if it arrives within your target (Sinch’s published handset target of 30 seconds is a useful default). An accepted API response alone is not a pass.
4. Log every probe with its timestamp, channel and result, and keep the vendor’s request IDs. Twilio wants the times from its status page plus “any documentation” of the unavailability. Sinch asks for relevant Call-IDs “within 7 days of issue” when you log a fault.
5. Compare your ledger with the status page each month. Where your failures line up with a posted incident, you have the evidence for a credit request (Twilio, Telnyx) or a fault record (Sinch). Where they do not, the problem is usually a carrier, a number or your own stack.

Monthly SLA	Failed probes allowed, 5-minute interval (8,640 probes in 30 days)	Failed probes allowed, 1-minute interval (43,200 probes in 30 days)	What to do when you pass it
99.9%	8.64	43.2	Past this, the channel ran below 99.9% for the month; review routing and consider a second carrier
99.95%	4.32 (the fifth failure breaks it)	21.6	Check which failures match a posted incident; on Twilio, file the claim within 30 days of month end
99.99%	0.864 (one failure breaks it)	4.32	At five-minute probes you cannot see 99.99% reliably; probe every minute if you are paying for it
99.999%	0.0864	0.432	Not measurable with one probe stream a month; no written voice or SMS SLA among the eight offers it

We tested the counting step on a synthetic 30-day probe log (8,640 voice and 8,640 SMS probes at five-minute intervals). The log held five consecutive failed voice probes and three scattered failed SMS probes. This short Python script read it:

import csv, sys
from collections import defaultdict
total, failed = defaultdict(int), defaultdict(int)
with open(sys.argv[1]) as f:
    for row in csv.DictReader(f):          # columns: timestamp,channel,ok
        total[row["channel"]] += 1
        failed[row["channel"]] += row["ok"] != "1"
for ch in sorted(total):
    up = 100 * (total[ch] - failed[ch]) / total[ch]
    print(f"{ch}: {total[ch]} probes, {failed[ch]} failed, measured uptime {up:.3f}%")

It printed “sms: 8640 probes, 3 failed, measured uptime 99.965%” and “voice: 8640 probes, 5 failed, measured uptime 99.942%”. Voice failed the four-probe rule and SMS passed it. Note the gap between ledgers. Those five voice failures were one 25-minute outage, which Twilio’s SLA would count because it ran longer than five continuous minutes. Five separate four-minute blips could also fail your rule if probes land in them (a five-minute probe can miss a four-minute blip), yet each would fall under Twilio’s five-minute exclusion. Sinch counts from one continuous minute; Telnyx’s file states no minimum.

Why voice and SMS uptime should be scored separately

A workflow that needs both channels, such as an AI receptionist that answers a call and then texts a booking confirmation, only works when both are up. If the two channels failed independently, both at 99.95%, the workflow would be up 0.9995 × 0.9995 = 99.90% of the time, about 43.2 minutes of exposure in a 30-day month instead of 21.6. They do not always fail independently. Under one vendor, a platform-wide incident can take out both at once, so read the real figure from your two ledgers rather than from the multiplication.

A second carrier is the usual answer to a budget you cannot live with, and the reasoning differs by channel. A dropped call is lost unless you retry or fail over. A text that fails at the API can usually be re-sent through a second provider, as long as your sender (10DLC campaign, toll-free number or registered sender ID) is also approved there. Our guide to building failover into an AI voice agent covers the voice side step by step.

Status pages: what each vendor shows for voice and SMS

A status page is the evidence base for any claim and the quickest way to tell a carrier problem from your own. We opened every vendor’s public status page and its incident-history page on 6 November 2026. We recorded what each one shows using the same three checks: separate voice and SMS components, a network or connectivity group (in the vendor’s own label), and a 90-day uptime display.

Platform	Status page	Separate voice and SMS components	Network or connectivity group (vendor’s label)	90-day uptime display	Incident history page
Twilio	status.twilio.com	Yes: Programmable Messaging; Programmable Voice per region (US1, AU1, IE1)	“Carrier Network”: 22 components, SMS, MMS and voice by region	Yes (“Uptime over the past 90 days”)	Yes
Sinch	status.sinch.com	Yes: Voice; Messaging APIs (27 components by protocol and region)	“External Connectivity”: 13 components, SMS and MMS only	Not shown on the page we saved	Yes
Vonage	vonageapi.statuspage.io (linked from Vonage’s system-status page)	Yes: SMS & Messages; Voice	“Global Connectivity”: 20 components, SMS and voice by region plus WhatsApp, Viber and Messenger	Yes	Yes
Plivo	status.plivo.com	Yes: Messaging API; Voice API; SIP Trunking	“Carrier Partners”: separate groups for Messaging API (9), Voice API (6) and SIP Trunking	Yes	Yes
Bandwidth	status.bandwidth.com	Yes: US messaging; US inbound and outbound calling; per-country calling groups	“International A2P Messaging Regional Connectivity”: 4 regions, messaging only	Not shown on the page we saved	Yes
Telnyx	status.telnyx.com	Yes: Messaging (inbound, outbound, long codes, toll-free, short codes, alphanumeric sender ID, 2FA); Voice API; SIP Trunking	Not seen among the components we saved	Not shown (page setting “display_uptime_mode”: “nothing”)	Yes
Infobip	status.infobip.com	Yes: SMS (19 locations); Voice (20 locations)	A separate page, external-connectivity-status.infobip.com, for operator and MNO issues by country and channel	Not shown on the page we saved	Yes
Bird	status.bird.com	Yes: SMS API and regional SMS connectivity; Voice API and voice connectivity	“Connectivity Platform”: 23 components, including SMS connectivity for APAC, EMEA, LATAM and NORAM and global voice connectivity	Yes	Yes

Every one of the eight runs a public status page that splits voice from SMS and keeps an incident history, so status transparency does not separate them much. The difference that matters for SMS is whether carrier-side trouble is posted. Twilio, Plivo, Vonage and Bird post carrier or connectivity components for both channels. Sinch and Bandwidth post them for messaging. Infobip runs a whole separate page for operator issues. Subscribe to the components you actually use, not the whole page.

Who each option is right and wrong for

Option	Right for	Wrong for
Zian AI (agent layer)	Teams that want an AI receptionist and outbound follow-up across voice, SMS, email, WhatsApp, Instagram and Messenger without building on raw carrier APIs	Anyone who needs a published uptime SLA from the agent vendor itself; zian.ai publishes none as at 6 November 2026
LeadsNow.ai (done for you)	Operators who want appointments set on a pay-per-result basis, with no carrier or agent to run	Teams that want to own and operate the telephony stack
Twilio	Buyers who want one percentage and a credit for both voice and SMS in a dated legal SLA before signing, and contact centres on Flex wanting 99.99%	Buyers counting on short blips: outages under five continuous minutes do not count, and the API credit is 10% against future invoices
Telnyx	Buyers who want a 99.99% voice and messaging figure with a three-tier credit schedule, with no plan or edition named in the file	Buyers who need that figure in their signed terms; the standard terms do not mention it, so get it written into a Service Order
Sinch	Buyers who want a written 99.95% figure plus published SMS latency targets	Buyers who need a financial remedy; the document has no credit
Vonage	High-volume buyers already paying for Premium or Enterprise support who want a 99.99% availability commitment	Buyers on Standard or Essentials support; the 99.99% line is not on those plans
Plivo	Buyers comfortable taking its stated “99.99% uptime SLA” and confirming the terms in their contract	Buyers who need the percentage and credits in a public SLA document; its SLA page states neither
Bandwidth	Buyers prepared to negotiate an SLA into an order form	Buyers who need a public, standard SLA to read before signing
Infobip, Bird	Buyers who will contract terms directly, and (Infobip) want operator-level status by country	Buyers who need a current published uptime percentage

For the voice-AI layers that sit on top of these carriers, such as speech, model and agent platforms, see what a reasonable enterprise SLA for voice AI uptime looks like. For carrier per-minute rates and a worked service-credit example, see voice network costs and the SLAs carriers sign.

When this answer changes

Three things would reorder the table. First, your contract: an enterprise order form with any of the eight can add a percentage and credits that no public page shows. Second, a vendor moving a figure into its legal terms. Telnyx’s 99.99% file, if its terms or SLA page adopted it, would rank first under our rule. Plivo already calls 99.99% an SLA on its own pages, and its llms.txt describes its SLA page as “Uptime commitments and service credit terms”. If that page printed the percentage and credits, Plivo would move up as well. Third, your own ledger. If one platform fails the four-probe rule on your numbers month after month, its written figure is no longer the number that matters to you.

Frequently asked questions

Which CPaaS has the best uptime SLA for voice and SMS?

On published terms read on 6 November 2026, Twilio and Telnyx. Twilio’s legal API SLA covers voice and messaging at 99.95% a month (99.99% with Enterprise Edition) with a 10% credit. Telnyx’s machine-readable SLA file states 99.99% for core voice and messaging with 10%, 25% and 50% credits, but its standard terms do not mention it.

Does Telnyx have an uptime SLA?

Telnyx publishes a machine-readable SLA file, telnyx.com/ai/sla.json, stating a “99.99% uptime SLA for core voice and messaging services” with credits of 10%, 25% and 50%. Its Terms and Conditions of Service do not mention an SLA, so ask for it in your Service Order.

How much downtime does 99.95% uptime allow in a month?

21.6 minutes in a 30-day month: (1 − 0.9995) × 43,200 minutes. In a 31-day month it is 22.32 minutes. Averaged across a year’s months it is about 21.9 minutes.

Does an SMS SLA guarantee my texts get delivered?

No. Twilio’s SLA excludes “telecommunications provider-related problems or issues”, and Sinch’s SLA excludes failures in networks outside where its service is hosted, such as “Service Providers or internet”. Both measure the platform, not the handset.

Does Vonage have an uptime SLA for its APIs?

Vonage’s API support-plans page lists a “99.99% platform availability SLA” on its Premium and Enterprise plans, not on Standard or Essentials. We could not find the SLA document behind it or a credit schedule, as at 6 November 2026.

Which carriers publish no uptime SLA?

Among the eight we read, we found no current published uptime SLA for Bandwidth, Infobip or Bird, as at 6 November 2026. Bandwidth prints 99.999% core network uptime on marketing pages, and MessageBird’s 99.95% SLA applied before 28 February 2022. Plivo calls 99.99% an SLA on its pages, but its SLA document states no percentage or credit.

How do I measure my own voice and SMS uptime?

Send one test call and one test SMS every five minutes to numbers you control on another network, and log each result. That is 8,640 probes per channel in a 30-day month; more than four failures on one channel means it ran below 99.95% for you.

Does Zian publish an uptime SLA?

No. zian.ai does not publish an uptime SLA or name its carrier, as at 6 November 2026. Zian is the AI agent layer that answers calls and follows up by voice, SMS and other channels on top of a carrier.

Where every figure on this page comes from

Figure	Who published it	Link	Date read
99.95% / 99.99% (Enterprise Edition) monthly API availability; 10% credit; 5-minute exclusion; 24-hour maintenance notice; 30-day claim window; carrier-problem exclusion	Twilio	Twilio APIs Service Level Agreement	6 November 2026
Flex: 99.99% for six Flex Critical Scenarios; 5% / 15% / 100% credits of Flex access fees; 5-minute intervals; carrier-problem exclusion	Twilio	Flex Critical Scenarios Service Level Agreement	6 November 2026
“99.99% uptime SLA for core voice and messaging services”; 10% / 25% / 50% credits below 99.99% / 99.9% / 99.0%; monthly formula; four exclusions incl. 72-hour maintenance notice; “proven”, “Historical uptime monitoring; third-party status page”	Telnyx	telnyx.com/ai/sla.json, linked from telnyx.com/llms.txt; earliest Wayback capture 4 September 2026	6 November 2026
“cannot guarantee … availability or performance” (section 11.11); no SLA mentioned	Telnyx	Terms and Conditions of Service	6 November 2026
“99.99% uptime” (WebRTC); “One platform, one SLA” (Voice AI Agents)	Telnyx	WebRTC, Voice AI Agents	6 November 2026
Minimum 99.95% availability (reasonable endeavours); 1-minute unavailability unit; SMS latency targets (10 seconds and 30 seconds for 90% of traffic); 10 business days’ maintenance notice; version 7.0 of 12 January 2024	Sinch	Messaging, Voice, and Verification SLA	6 November 2026
99.50% (reasonable endeavours), India Version 2, 22 July 2024	Sinch India	Messaging Service Level Agreement (India)	6 November 2026
“99.99% platform availability SLA” on Premium and Enterprise support plans	Vonage	API Support Plans (read via r.jina.ai; direct requests returned HTTP 403)	6 November 2026
“We offer a 99.99% uptime SLA”; “99.99% uptime SLA”; “99.99% Uptime SLA available”	Plivo	Twilio alternative page	6 November 2026
“a 99.99% platform uptime SLA” (site-wide structured data); /legal/sla/ described as “Uptime commitments and service credit terms”	Plivo	SLA page source, plivo.com/llms.txt	6 November 2026
No availability percentage or credit in the SLA document (last updated 20 December 2024)	Plivo	Platform Service Levels and Support Services – API	6 November 2026
“99.999% core network uptime”; “99.999% uptime” (All-IP core); “100% uptime for toll‑free traffic” (network page); Silver plan “SLA-backed responses and uptime”; “any agreed-upon service level agreement”	Bandwidth	Home page, Network, Customer Success Plans, Communications Services Agreement	6 November 2026
No guarantee the services “will be available to You at all times” (terms last updated 29 September 2026)	Infobip	Service Terms and Conditions	6 November 2026
Separate operator connectivity status page, by country and channel	Infobip	Operator connectivity status	6 November 2026
Unavailability remedies left to an Order Form (terms last updated 21 November 2024)	Bird	Terms of Service	6 November 2026
MessageBird 99.95% with service credits, “applied before 28 February 2022”	Bird	Archived MessageBird SLA	6 November 2026
Status-page component counts (Twilio Carrier Network 22; Sinch External Connectivity 13, Messaging APIs 27; Vonage Global Connectivity 20; Plivo Carrier Partners 9 and 6; Bandwidth 4; Infobip SMS 19, Voice 20; Bird Connectivity Platform 23)	Each vendor’s status page	Twilio, Sinch, Vonage, Plivo, Bandwidth, Telnyx, Infobip, Bird	6 November 2026
Downtime minutes, failed-probe allowances, 99.90% combined availability, synthetic probe-log results	Zian AI (our arithmetic, formulas shown on this page)	This page	6 November 2026