Resemble AI's pricing now covers deepfake detection and media security only. In an August 13, 2026 post, Resemble said it is not selling voice AI to new customers. It supports existing voice customers but is not taking new ones. The official pricing page lists no text-to-speech, voice cloning, or speech-to-speech rates. If you came here to budget for voice generation, the answer is that Resemble no longer sells it to new buyers.
What you can buy is a detection stack: Detect for audio, images, and video; Intelligence for plain-language explanations of a result; Identity search; Watermarker; Agent Detection; and Meetings protection. Every plan includes platform and API access.
Meter | Flex | Team and Business |
|---|
Detect, audio | $0.035 per second | $0.015 per second |
Detect, images | $0.035 per image | $0.015 per image |
Detect, video | $0.07 per second | $0.03 per second |
Intelligence | $0.025 per second | $0.015 per second |
Per-call meters cost the same on Flex, Team, and Business: Identity search costs $0.0005 per call, Watermarker encode $0.0005 and decode $0.0002 per call, and Agent Detection $1 per 1,000 sessions. Enterprise offers volume pricing on every meter.
The rates show where costs grow. Audio and video are billed by the second, so a contact center screening calls pays by call length, not call count. On Flex, one hour of audio detection costs $126 (3,600 seconds × $0.035), and the same hour at the Team rate costs $54. One hour of video costs $252 on Flex and $108 on the plan rate. Identity, watermarking, and agent detection are cheap per call and rarely drive the bill.
Start on Flex. Load a small credit balance, run your real files through Detect and Intelligence, and measure accuracy, latency, and how much analyst review each result needs. Credits never expire, so a pilot does not waste money if it pauses.
Move to Team when two things are true: several people need seats, and monthly detection volume is high enough for the lower rates to cover the subscription. With audio detection, the Team subscription ($280 a month billed annually) is covered by the rate saving at about 3.9 hours of audio a month (280 ÷ 0.02 per second ≈ 14,000 seconds). Below that, Flex is cheaper unless you need the extra seats, batch uploads, or large-file uploads.
Choose Business for SSO, 20 seats, and organization-wide Meetings protection. Resemble suggests talking to sales once Business-plan spend passes about $1,000 a month, since volume pricing can be cheaper. Go straight to Enterprise if you need SLAs, on-premises deployment, custom model training, or SOC 2 documentation before production.
New customers cannot buy hosted voice generation, cloning, speech-to-speech, or voice agents from Resemble, according to its strategy post. Existing voice customers should confirm their contract terms and renewal path directly with Resemble.
The voice models remain available as open source. Resemble's former text-to-speech page now redirects to a product notice saying voice generation is no longer offered as a commercial product, and its Chatterbox page says Chatterbox and its other voice models stay open and free to build on but are no longer sold as paid products. That page describes Chatterbox, Chatterbox Multilingual, and Chatterbox Turbo as MIT-licensed on GitHub and Hugging Face, with cloning from about 5 seconds of reference audio and 23 or more languages. Self-hosting has no license fee, but you pay for compute and own hosting, consent checks, abuse prevention, and support. The same page's FAQ still mentions managed hosting at app.resemble.ai, which conflicts with its own notice. Confirm with Resemble before you budget for it.
For a paid, hosted voice service, compare ElevenLabs, Cartesia, and Fish Audio. Many teams will pair one of those for generation with Resemble Detect for verification.
