Voice Agent Pricing 2026: What Vapi, Retell, Bland & ElevenLabs Really Cost/Min

What Voice Agents Actually Cost Per Minute

A typical AI voice agent runs about $0.08 to $0.14 per minute all-in — including text-to-speech, speech-to-text, the LLM and US telephony. The four leading platforms (Vapi, Retell AI, Bland AI and ElevenLabs Agents) have converged into that band, but they get there through completely different billing models, and the real cost drivers are your provider choices, monthly platform fees and per-number or per-concurrency charges — not the headline per-minute rate.

That's the trap. Vapi advertises $0.05/min, Retell $0.07/min, Bland $0.14/min and ElevenLabs $0.08/min — but those numbers aren't measuring the same thing. Some are orchestration fees on top of provider costs; some bundle everything in. This guide lines them up on the same basis using each platform's current pricing page (checked September 2026), so you can compare like for like.

Voice Agent Platforms Compared

"Typical all-in" models a standard-quality TTS voice, a mid-tier LLM and US telephony. Figures are built from each platform's published component rates (verified September 2026), not a billed account.

Vapi: Orchestration Fee + At-Cost Providers

Vapi charges a $0.05/min orchestration fee and passes every provider through at cost with no markup. Its pricing page lists the ranges: Deepgram transcription at $0.0095–$0.0099/min, an OpenAI LLM at $0.0077–$0.0452/min, and ElevenLabs voice at $0.0146–$0.0238/min. Internal transport is free; Twilio, Vonage and Telnyx telephony is extra. Vapi's own worked example puts 1,000 minutes at roughly $82–$129/month (before external telephony) — which, notably, makes it one of the cheapestoptions at typical volumes, not the most expensive.

The catch is operational, not financial: you manage the keys and the integrations. Success Packages ($0 to start, $29/month Core, $999/month Pro, custom Premier) gate support and a hosting-fee discount — the Pro package cuts the orchestration fee to 10% of hosting, which starts paying off above roughly 20,000 minutes a month.

Retell AI: One Rate, Toggleable Components

Retell's $0.07–$0.31/min range is the sum of parts you switch on: $0.055/min for voice infrastructure, $0.015/min for standard TTS voices (Minimax, Fish, Cartesia, OpenAI, Inworld) or $0.040/min for ElevenLabs voices, an LLM at $0.0016–$0.064/min depending on model, and US telephony at $0.015/min. Add-ons — knowledge base (+$0.005/min), advanced denoising (+$0.005/min), PII removal (+$0.01/min) — stack on top, and AI quality assurance is $0.10/min after the first 100 free minutes.

A lean build (standard voice, cheap LLM, US telephony) lands near $0.11/min; loading ElevenLabs voices, a frontier LLM and QA pushes toward the $0.31 ceiling. Operational costs are itemized too: concurrency beyond 20 calls is $8/slot/month, extra knowledge bases $8/month, and verified phone numbers $10/month. See the full Vapi vs Retell AI comparison for a head-to-head.

Bland AI: Flat, Fully Bundled

Bland folds LLM, speech-to-text and text-to-speech into a single per-minute rate with no separate token charges — the simplest billing of the four. The Start plan is $0.14/min with no monthly fee (and includes 2 credits plus an inbound number worth $15/month); the Build plan is $0.12/min for a $299/month platform fee. Transfers cost $0.05/min (Start) or $0.04/min (Build), and enterprise pricing is contracted to volume. It's purpose-built for high-volume outbound, where a predictable flat rate beats modeling a compounding stack — the trade-off is less provider flexibility than Vapi.

ElevenLabs Agents: Voice-Native, Plan-Included Minutes

ElevenLabs Agents bills additional call minutes at $0.08/min (doubling to a $0.16/min burst rate when you exceed your workspace concurrency), with text messages at $0.003 each. Crucially, minutes are included in the subscription tiers: Free covers 15 minutes/month, Starter 75 ($6/month), Creator 275 ($22/month), Pro 1,238 ($99/month), Scale 3,738 ($299/month) and Business 12,375 ($990/month). Voice and LLM are bundled into the agent minute; you add telephony via Twilio or SIP. For quality-driven use cases — sales calls where rapport matters — ElevenLabs' voice is the reason to pay the premium, and if your volume fits inside a plan's included minutes it can be the cheapest option outright.

Monthly Cost at Different Volumes

Estimates from published rates (September 2026). ElevenLabs at 10K minutes assumes the Business plan (12,375 included minutes) or Scale plan plus $0.08/min overage. Vapi's Pro package changes its math above ~20K minutes. Run your own numbers in the TTS cost calculator.

Hidden Costs to Budget For

• Telephony: a per-minute phone charge applies on top on Vapi and ElevenLabs (billed by your carrier), and Retell itemizes US telephony at $0.015/min — more internationally.
• LLM inference: On unbundled platforms this is a separate line item that can rival the TTS cost — a frontier model minute costs far more than a small one.
• Concurrency & numbers: Retell charges $8/concurrency slot and $10/verified number monthly; budget these before you scale past a handful of simultaneous calls.
• Burst pricing: ElevenLabs doubles to $0.16/min when you exceed your plan's concurrency limit.
• Self-hosting the TTS: running an open model like Kokoro can cut the voice line item to near zero if you have the GPU and engineering to support it.

Which Platform to Choose

• Lowest cost + control: Vapi.
• Predictable single invoice: Retell AI.
• High-volume outbound: Bland AI.
• Best voice quality / plan-included minutes: ElevenLabs Agents.

Whichever platform you pick, the TTS engine inside it drives both latency and a big share of the cost — so choose that deliberately. Our best TTS for voice agents guide covers the voice layer, and the full TTS pricing comparison lines up every provider on a per-character and per-minute basis.

Related Guides