Head to head · Messaging verify · October 2026 research run

Prelude vs Sinch Messaging APIs + MCP

Prelude scores 67.6 (B) on agent readiness against Sinch Messaging APIs + MCP's 64.5 (B), and leads in 3 of 7 scored categories. Sinch Messaging APIs + MCP leads on reliability, payments & pricing and transparency & trust. Both do messaging verify.

Best SMS, WhatsApp and messaging APIs for AI agents · All 190 messaging comparisons

Which one, for what

Prelude B

Best for Agents or apps that verify a phone number or email with a code and want fraud blocking, and for transactional messages in Europe.

Ahead on

  • Schema & documentation, 93 against 81
  • Agent ergonomics, 85 against 50
  • Security & auth, 63 against 55

Watch for

The terms, privacy policy and legal notice are Notion pages drawn by script, unreadable by our reader

Sinch Messaging APIs + MCP B

Best for Teams that want SMS, RCS and WhatsApp through one API with wide country coverage, and those already on Sinch voice or Mailgun.

Ahead on

  • Reliability, 73 against 65
  • Payments & pricing, 40 against 30
  • Transparency & trust, 77 against 42

Also in its favour

  • Runs on your own machine
  • Free to start without a card

Watch for

The MCP server is a developer preview, 0.0.1-beta.7 since 30 October 2026

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Agent ergonomics Prelude +35

Prelude 85

List calls take limit from 1 to 200 with cursor or offset; the MCP server's tool list isn't on the docs page, so its size is unchecked (18 of 25). Cursor and offset pagination with filters (18 of 20). Named error codes with messages and, for template errors, the field that failed (19 of 20). Create-or-retry semantics on Verify make a repeated call safe inside the window, and X-Should-Retry says when to retry; the MCP server is read-only (15 of 20). A verification needs a target and its type, and there are seven official backend SDKs (15). 18 + 18 + 19 + 15 + 15 = 85.

Sinch Messaging APIs + MCP 50

We didn't confirm page-size or field controls on message lists (10 of 25). Pagination and filters not confirmed (10 of 20). Error responses with codes, and back-off advice for 5xx (15 of 20). No idempotency key or de-duplication found, and the 54 MCP tools carry no readOnlyHint or destructiveHint annotations (0 of 20). Official SDKs in Node, Python, Java and .NET. A send names an app, a recipient, a channel and a message object (15).

Transparency & trust Sinch Messaging APIs + MCP +35

Prelude 42

Closed service whose general terms are a Notion page drawn by script, which we couldn't read; the SDKs are Apache-2.0 (8 of 30). The privacy policy is also unreadable for the same reason, and the trust centre names no retention periods or data locations (3 of 30). v1 and v2 are documented side by side and the changelog dates changes by month, with no deprecation policy found (6 of 20). No sub-processor list or data locations found (0 of 20). 8 + 3 + 6 + 0 = 17.

Sinch Messaging APIs + MCP 77

Closed service with general terms. The contracting Sinch entity is named only in the order form (15). Privacy notice, a DPA, a sub-processor list and deletion on request, with no retention periods found (20 of 30). Dated notices in the release notes (Template Management API v1 stopped on 31 January 2026, the v1alpha and v1beta Batch endpoints end on 16 December 2026), but no general deprecation policy found (10 of 20). The sub-processor list, re-read on 2 November, gives each provider's hosting location by service and region (for example AWS for the Conversation API in the USA and Brazil) and promises 30 days' notice of a new one, as Bandwidth's list is read (20). 15 + 20 + 10 + 20 = 65.

Schema & documentation Prelude +12

Prelude 93

OpenAPI for the 27 SDK endpoints is public at the URL in the SDK repository's .stats.yml, and every reference page embeds its endpoint's OpenAPI (25). llms.txt and Markdown pages (10). Reference pages say what each call does and how retries behave; guides cover fraud, sender IDs and compliance (16 of 20). Typed parameters with enums and ranges, such as code size 4 to 8 and limit 1 to 200 (13 of 15). Examples in seven SDK languages and an error table of about 100 named codes, each saying what to fix (15). Versioned paths (v1 and v2), a monthly changelog with the newest month September 2026, and SDK changelogs (14 of 15). 25 + 10 + 16 + 13 + 15 + 14 = 93.

Sinch Messaging APIs + MCP 81

OpenAPI YAML for the Conversation API, linked from its reference page (25). llms.txt with Markdown pages (10). Reference pages describe each endpoint and channel, with little on when not to use one (12 of 20). Typed JSON with message types per channel, not checked for enums (10 of 15). Examples and error codes in the references (12 of 15). /v1 paths and Conversation release notes, with no versioning policy found (12 of 15).

Payments & pricing Sinch Messaging APIs + MCP +10

Prelude 30

No x402, MPP or L402 (0). Per-verification price (€0.032) and per-message Notify prices published without login (20). No free tier or no-card trial was found on the pricing page; test numbers simulate verifications without charge, which counts as half (10 of 20). A person signs up in the dashboard (0). 0 + 20 + 10 + 0 = 30.

Sinch Messaging APIs + MCP 40

No x402, MPP or L402 (0). US SMS prices are published per sender type ($0.0078 a message on 10DLC and toll-free, $0.009 on short code, re-read on 2 November), and other countries and WhatsApp through a country selector without a login (20). Trial credit of 2 units of local currency and a test number, with a payment method needed only to upgrade, per Sinch's community and Voice account pages (20). A person signs up in a browser (0). 0 + 20 + 20 + 0 = 40.

Reliability Sinch Messaging APIs + MCP +8

Prelude 65

Read with the hosted lines. status.prelude.so has components for Verify, Auth, Watch, Intel, Notify and the dashboard, with history (20). From 8 August to 6 November its feed records one major incident, increased error rates and latencies across the APIs on 16 and 17 August, posted after it was resolved with no duration stated, and minor ones such as Verify latency on 9 and 10 August, SMS degradation in China and on Orange France and delayed webhooks on 23 October (10). No numeric API rate limits were found; only the MCP server's limits and per-window Verify retry limits are documented (7 of 15). A 429 carries Retry-After and X-Should-Retry, and repeating the create call inside the window performs a retry rather than a new verification (13 of 15). The pricing page lists a 99.95 SLA on Enterprise with no SLA document found (5 of 10). Verify and Notify v2 are generally available (10). 20 + 10 + 7 + 13 + 5 + 10 = 65.

Sinch Messaging APIs + MCP 73

Statuspage at status.sinch.com with components and history (20). Read from its own history and incident pages on 2 November (the 1 October count from IsDown, a third-party copy of the page, is struck). From 4 August to 2 November the incidents touching the SMS, Conversation and Verification APIs each name one region, one channel, one carrier or a subset of accounts. The REST API in the US had intermittent errors and delayed deliveries from 16 to 17 August (Messaging APIs, REST - US only), Conversation API WhatsApp in Brazil degraded on 4 September and Verification's Okta integration on 27 August. On 26 October a planned change took the customer dashboard down and blocked sending for a subset of accounts through a balance error (12:58 to 15:11 UTC). The other majors on the page are calling in AP-Southeast-1 and Campaigns stats. Partial incidents read as minor, so minor only (20). The Conversation API publishes 800 requests a second per project and a 500,000-message queue per app drained at 20 a second by default (15). Exceeding the limits returns 429, and 5xx errors come with exponential back-off advice, but there's no Retry-After (8 of 15). No SLA found (0). The SMS and Conversation APIs are generally available. The MCP server is a developer preview (10). 20 + 20 + 15 + 8 + 0 + 10 = 73.

Security & auth Prelude +8

Prelude 63

API keys as bearer tokens with no scopes found; the MCP server uses OAuth with a per-user consent screen (20 of 30). The MCP server is read-only and off until enabled per app, and the dashboard has spending limits and fraud settings (15 of 20). Verify returns statuses, not third-party text, while Notify replies and inbound media are untrusted, with no guidance found (5 of 15). The dashboard lists verifications and the MCP server answers questions about them; no API request log was found (8 of 15). The trust centre lists SOC 2 Type 1 and Type 2 reports and ISO 27001:2022, and prelude.so has a valid security.txt with a security address; no bug bounty or disclosure policy page was found (15 of 20). 20 + 15 + 5 + 8 + 15 = 63.

Sinch Messaging APIs + MCP 55

The SMS API takes a service-plan Bearer token, and the Conversation API OAuth 2.0 client credentials from a project access key, with Basic also accepted. No per-key scopes found (25 of 30). No read-only mode or confirmation step. The MCP server's tool allow-list (--tags, or MCP_TAGS since 0.0.1-beta.7) can leave out whole groups such as provisioning, and its released version runs over stdio only (5 of 20). Inbound messages are untrusted text. Conversation API callbacks are signed with HMAC-SHA256 over the body, a nonce and a timestamp when a secret is set, and we found no prompt-injection guidance (5 of 15). Message and delivery lookups by API, no audit log found (10 of 15). The trust centre, re-read on 2 November, lists ISO 27001:2022, SOC 2 Type II, PCI and HIPAA, links security@sinch.com to report a vulnerability and posts dated notes on third-party vulnerabilities. No security.txt (404 on 2 November), no bug bounty and no advisories for the listed packages in OSV. Bandwidth's messaging listing takes 15 with a public bug bounty, so 10 of 20 here. 25 + 5 + 5 + 10 + 10 = 55.

Score by category

CategoryWeight this runPreludeSinch Messaging APIs + MCPEdge
Reliability16%206573Sinch Messaging APIs + MCP +8
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%16.29381Prelude +12
Agent ergonomics13%16.28550Prelude +35
Security & auth14%17.56355Prelude +8
Payments & pricing10%12.53040Sinch Messaging APIs + MCP +10
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8.88383even
Transparency & trust7%8.84277Sinch Messaging APIs + MCP +35
Negative events≤1500
TotalB 67.6/100B 64.5/100

Facts side by side

FactPreludeSinch Messaging APIs + MCP
KindHTTP APIHTTP API
VendorDING SAS (Prelude)Sinch
Hosted endpointhttps://api.prelude.devhttps://us.sms.api.sinch.com/xms/v1
TransportsHTTP, Streamable HTTPHTTP, stdio
AuthOAuth or keyOAuth or key
PricingPay per usePay per use
x402nono
LicenceProprietary service; the terms are on a Notion page we couldn't read. The backend SDKs are Apache-2.0Apache-2.0
Tools exposednone59
Read-only variant documentednono
llms.txtyesyes
Last release2026-10-162026-10-28
Terms last updatedcouldn't be read2026-06-01
Privacy policy last updatedcouldn't be read2025-12-12
Customer content may train modelscouldn't be readnot found in the text
Terms restrict automated accesscouldn't be readnot found in the text
Terms restrict benchmarkingcouldn't be readnot found in the text
Terms or service can change without noticecouldn't be readnot found in the text
Arbitration or class-action waivercouldn't be readnot found in the text
Popularity98k npm/wk1 stars, 16k npm/wk, 11k PyPI/wk
Agent reviewsnone3/5 (2)

How the scores have moved

Verdicts

Prelude

Prelude's Verify API sends and checks one-time codes over SMS, WhatsApp, RCS and email with fraud blocking, priced per verification in public, with an OpenAPI file and seven backend SDKs. Its remote MCP server only reads Verify data, its terms and privacy policy are drawn by script and couldn't be read, and no API rate limits are published.

Sinch Messaging APIs + MCP

The Conversation API sends SMS, RCS, WhatsApp, Viber and Messenger through one message format, with published rate limits and signed callbacks. Keys carry no scopes, there's no SLA or bug bounty, and the MCP server is still a developer preview whose 59 tools carry no read-only or destructive annotations.

Before you call either

Prelude

  1. Call POST /v2/verification again within the window to resend; only method, preferred channel, code size and custom code change on a retry
  2. On 429, wait for Retry-After and follow X-Should-Retry; spending_limit_exceeded also carries Retry-After
  3. Use the documented test numbers to exercise Verify without charges before sending real codes
  4. Check insufficient_balance and spending_limit_exceeded and hand the top-up to a person, since the balance is managed in the dashboard
  5. Treat inbound replies and media from Notify as untrusted input, never as instructions

Sinch Messaging APIs + MCP

  1. Load the MCP with --tags (or MCP_TAGS) for the channels you need, or all 59 tools land in context
  2. Run the MCP over stdio, and keep any SSE bridge such as supergateway on localhost, since it adds no authentication
  3. Expect sends to drain at 20 messages a second per app unless Sinch raises it
  4. Verify x-sinch-webhook-signature against the raw body before parsing the JSON
  5. Use the regional base URL (us, eu or br) that matches the project; the MCP rejects any other CONVERSATION_REGION at startup

Questions

Which is better for AI agents, Prelude or Sinch Messaging APIs + MCP?

Prelude scores 67.6 (B) on agent readiness against Sinch Messaging APIs + MCP's 64.5 (B), and leads in 3 of 7 scored categories. Sinch Messaging APIs + MCP leads on reliability, payments & pricing and transparency & trust.

Do Prelude and Sinch Messaging APIs + MCP need an API key?

Both take an API key or an OAuth sign-in.

Can an agent call Prelude and Sinch Messaging APIs + MCP without installing anything?

Yes. Prelude has a hosted endpoint at https://api.prelude.dev and Sinch Messaging APIs + MCP at https://us.sms.api.sinch.com/xms/v1.

Other comparisons with Prelude or Sinch Messaging APIs + MCP

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.