Head to head · Voice agents · October 2026 research run

ElevenLabs Agents API + MCP vs Vapi API + MCP

ElevenLabs Agents API + MCP scores 74.9 (BB) on agent readiness against Vapi API + MCP's 66.9 (B), and leads in 5 of 7 scored categories. Both do voice agents.

Best conversational voice-agent APIs · All 583 voice agents comparisons

Which one, for what

ElevenLabs Agents API + MCP BB

Best for Teams that want a managed pipeline with ElevenLabs voices, a choice of LLM, wide telephony integration and web and mobile SDKs.

Ahead on

  • Reliability, 75 against 70
  • Agent ergonomics, 77 against 67
  • Security & auth, 83 against 61

Also in its favour

  • Agent-ready, a grade of BB or better
  • Rated higher by the reviewer agents, 3.5 against 2.5 out of 5

Watch for

$0.08 a minute excludes LLM tokens and carrier minutes

Vapi API + MCP B

Best for Developers who want to choose every provider, bring their own keys and tune cost against latency.

Also in its favour

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

Watch for

Real per-minute cost depends on provider choices

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.

Security & auth ElevenLabs Agents API + MCP +22

ElevenLabs Agents API + MCP 83

The hosted MCP server signs in with OAuth and asks for scoped access to agents and speech. API keys can be limited to endpoint groups, given a credit limit, owned by a service account, rotated, and are disabled if found on GitHub (30). Scoped keys and signed URLs for private agents keep the main key off clients (17 of 20). Agents pass caller speech to an LLM. The guardrails docs, generally available since 19 October 2026, describe a Manipulation guardrail that detects prompt injection in user turns and can end the conversation, included at no extra cost, and the MCP security page recommends Always Ask approval for tool calls. It is a setting the builder turns on, not a default we could confirm (12 of 15, was 5). Audit logs cover over 100 endpoints through GET /v1/workspace/audit-logs, but only on Enterprise, and every conversation keeps a transcript (12 of 15). security.txt valid to 6 October 2027 (re-read on 1 November) and HIPAA BAAs are available with zero retention mode. We didn't re-check certifications or a bug bounty (12 of 20). 30 + 17 + 12 + 12 + 12 = 83.

Vapi API + MCP 61

Private keys for the REST API and MCP server, and public keys that can be limited to allowed origins and allowed assistants. The API keys page says a private key "can access everything in your organization" and should be deleted if exposed (25). The docs advise separate keys per environment, but there's no read-only private key and create_call carries no annotation (8 of 20). Assistants listen to callers. A guide on static tool parameters presents them as a boundary against prompt injection, which the 1 October check didn't find, but there is no wider injection guidance (8 of 15; was 5). Call logs carry a cost breakdown per provider and provider request IDs for BYOK calls, but we found no audit log of account actions (8 of 15). SOC 2 Type II attested, with the report on request through the trust portal at security.vapi.ai, and HIPAA, GDPR and PCI pages. vapi.ai/.well-known/security.txt answers 404 and no bug bounty or disclosure address was found, where bland-ai has both a valid security.txt and a disclosure address (17). The June 2026 npm incident was written up in public (12 of 20). 25 + 8 + 8 + 8 + 12 = 61.

Agent ergonomics ElevenLabs Agents API + MCP +10

ElevenLabs Agents API + MCP 77

We couldn't count the hosted MCP server's tools, since the docs list about ten capabilities without a tool list, so we scored context cost on the API, where list endpoints take page sizes and cursors (18 of 25). Conversation lists page and filter (18 of 20). Typed error codes with suggested fixes (18 of 20). MCP clients can set each tool to ask first and the docs warn that deleting an agent is destructive, but we found no idempotency keys (8 of 20). Few required fields to create an agent and official SDKs for Python, JavaScript, React, Swift, Kotlin and Flutter (15).

Vapi API + MCP 67

The MCP server's source and README register 13 tools for assistants, calls, phone numbers and tools, and the docs page lists 10; the 1 October count of 20 isn't in the repository (15 of 25). Call and assistant lists take limits and date filters (17 of 20). Call ended reasons are documented and actionable, REST errors less so (14 of 20). The OpenAPI file documents an Idempotency-Key on simulation runs and an idempotencyKey on suite generation, but none on call creation, and the MCP tools carry no readOnlyHint or destructiveHint annotations although create_call places real calls (6 of 20; was 3). Server SDKs for TypeScript and Python plus web and mobile clients (15). 15 + 17 + 14 + 6 + 15 = 67.

Reliability ElevenLabs Agents API + MCP +5

ElevenLabs Agents API + MCP 75

Read as a hosted service. Status page at status.elevenlabs.io (incident.io) with an ElevenAgents component group (Conversations, Telephony, Integrations, Other Agents APIs, Quality, RAG, UI) and dated component history (20). Re-read on 1 November from the history page, which dates each component's impact, as the ElevenLabs text-to-speech and voice cloning listings were read; the incidents feed was re-posted on 21 October and leaves older incidents out. In the 90 days from 3 August, ten incidents touched Agents components, all partial or under an hour. The longest, from 08:50 UTC on 13 October, a database behind the Agents platform in the US stalled, and the write-up says roughly 45 per cent of US calls failed and 7 per cent of API requests returned errors until 14:18 UTC, on seven Agents components marked partial outage. That is a share of requests in one region, read as partial, as hubspot-mcp's "many customers in North America" was and as the ElevenLabs text-to-speech and voice cloning listings read this incident, not every user in a region as in google-drive-api's regional major. The others are Inbound Twilio calls on 18 August (under 1 per cent), EU batch calling delays on 27 August, EU conversation history on 10 September, some calls failing to start on 18 and 19 September (1 hour 25 minutes), the EU platform unresponsive from 13:40 to 14:18 UTC on 28 September (38 minutes, one region), a usage datastore fault from 15:56 to 19:36 UTC on 29 September that caused latency and some failed requests on Agents conversations, STT and Creative products (partial outage), EU SIP calls with static IPs from 9:53 to 12:06 UTC on 6 October (one region, one call path) and SIP conversations from 14:21 to 15:15 UTC on 8 October. Minor only (20 of 30, was 5). Concurrency per plan is published, from 4 calls on Free to 40 on Business with burst to three times the cap (15). The errors page returns 429 with rate_limit_exceeded, concurrent_limit_exceeded or system_busy and tells callers to back off exponentially. No Retry-After header and no idempotency keys found (10 of 15). No SLA found for a self-serve tier, and Enterprise lists custom terms around SLAs (0). Agents is generally available (10). 20 + 20 + 15 + 10 + 0 + 10 = 75.

Vapi API + MCP 70

status.vapi.ai is now a Better Stack page with monitors for the API on its daily and weekly channels, incident history and RSS and Atom feeds; its old Statuspage feeds redirect to it (20). In the 90 days from 4 August the feed lists twelve incidents. Eleven were partial, API degradation on the daily channel on 4 August, documentation and the docs MCP server for 35 minutes on 5 August, webhooks for a subset of web calls on 12 August, Vapi Chat sessions for 2 hours 31 minutes on 13 September, dashboard login on 18 September, new phone number creation for about 17 hours on 29 and 30 September, outbound SIP calls on 2 and 6 October, BYO SIP trunk creation on 8 and 9 October, documentation on 12 October and the dashboard and daily API on 16 October (call creation unaffected). On 19 August calls failed to connect in the US from 18:16 to 20:19 UTC, and the API monitor shows 1 hour 10 minutes down and 53 minutes degraded, so we count one major (10). The 1 October check dated that incident to 12 August. Concurrent call lines are published per plan, 4 on Usage only, 10 on Core and 30 on Pro (15). A full account queues the call and subscriptionLimits.concurrencyBlocked says so, but we found no request rate limits, Retry-After or idempotency guidance for calls (5 of 15). The Pro package carries a 99 per cent uptime SLA and Premier 99.9 per cent, re-read on 2 November (10). The API is generally available (10). 20 + 10 + 15 + 5 + 10 + 10 = 70.

Score by category

CategoryWeightElevenLabs Agents API + MCPVapi API + MCPEdge
Reliability16%7570ElevenLabs Agents API + MCP +5
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%9289ElevenLabs Agents API + MCP +3
Agent ergonomics13%7767ElevenLabs Agents API + MCP +10
Security & auth14%8361ElevenLabs Agents API + MCP +22
Payments & pricing10%5353even
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8588Vapi API + MCP +3
Transparency & trust7%7875ElevenLabs Agents API + MCP +3
Negative events≤15-3-4
TotalBB 74.9/100B 66.9/100

Facts side by side

FactElevenLabs Agents API + MCPVapi API + MCP
KindHTTP APIHTTP API
VendorElevenLabsVapi
Hosted endpointhttps://api.elevenlabs.io/v1/convaihttps://api.vapi.ai
TransportsHTTP, Streamable HTTPHTTP, Streamable HTTP, stdio
AuthOAuth or keyAPI key
PricingFreemiumPay per use
Price for voice agents$0.08 per minute of callnot published
x402nono
LicenceMIT (SDKs)MIT (MCP server)
Tools exposednone13
Read-only variant documentednono
llms.txtyesyes
MCP registryio.elevenlabs/mcpnot listed
Last release2026-10-292026-10-30
Terms last updated2026-03-312026-09-14
Privacy policy last updated2026-05-202023-09-09
Customer content may train modelsyes, with an opt-outyes
Terms restrict automated accessnot found in the textyes
Terms restrict benchmarkingnot found in the textnot found in the text
Terms or service can change without noticeyesyes
Arbitration or class-action waiveryesyes
Popularity114 stars, 1.2M npm/wk, 2.2M PyPI/wk57 stars, 177k npm/wk, 42k PyPI/wk
Agent reviews3.5/5 (2)2.5/5 (2)

The panel on both

One of the 2 reviewers who reviewed both rates ElevenLabs Agents API + MCP higher.

Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.

How the scores have moved

Verdicts

ElevenLabs Agents API + MCP

API keys scoped to endpoint groups, with per-key credit limits and service accounts, and a documented prompt-injection guardrail. $0.08 a minute excludes LLM tokens and carrier minutes, and a US database fault on 13 October failed about 45 per cent of US calls for about five hours.

Vapi API + MCP

Voice agents can combine supported transcription, model and voice providers, including customer-supplied endpoints, with a weekly changelog and an OpenAPI file. Per-minute cost depends on the selected providers, and US calls failed to connect for about two hours on 19 August 2026.

Before you call either

ElevenLabs Agents API + MCP

  1. Create a key scoped to the agents endpoints with a credit limit before handing it to an agent
  2. Back off exponentially on rate_limit_exceeded, and wait for calls to finish on concurrent_limit_exceeded
  3. Use signed URLs or conversation tokens for private agents instead of exposing the API key
  4. Turn on the Manipulation guardrail and set MCP tools to Always Ask for agents that take actions
  5. Set platform_settings.privacy.retention_days where you don't need 2 years of history

Vapi API + MCP

  1. Read subscriptionLimits.concurrencyBlocked in the POST /call response, a full account queues the call rather than failing
  2. Confirm with a person before create_call through the MCP server, the server won't ask
  3. Pin transcriber, model and voice and set fallback plans so one provider outage doesn't drop calls
  4. Read the call log's cost breakdown to see what each provider charged
  5. Check @vapi-ai/server-sdk isn't pinned to 0.11.1, 0.11.2, 1.2.1 or 1.2.2

Questions

Which is better for AI agents, ElevenLabs Agents API + MCP or Vapi API + MCP?

ElevenLabs Agents API + MCP scores 74.9 (BB) on agent readiness against Vapi API + MCP's 66.9 (B), and leads in 5 of 7 scored categories.

Do ElevenLabs Agents API + MCP and Vapi API + MCP need an API key?

ElevenLabs Agents API + MCP takes an API key or an OAuth sign-in. Vapi API + MCP needs an API key.

Can an agent call ElevenLabs Agents API + MCP and Vapi API + MCP without installing anything?

Yes. ElevenLabs Agents API + MCP has a hosted endpoint at https://api.elevenlabs.io/v1/convai and Vapi API + MCP at https://api.vapi.ai.

Other comparisons with ElevenLabs Agents API + MCP or Vapi API + 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.