Resemble AI review

Resemble AI is a deepfake detection platform for audio, video, and images. The company says its in-house DETECT-World model flags generated media in real time. Teams can use it through an API, SDK, or native integrations such as Microsoft Teams, where detection runs inside meetings without a bot. Developers can install the Python package with pip install resemble; the REST API returns a deterministic detection score from a single call.

The platform also offers voice and likeness identity matching, along with PerTh multimodal watermarking, which Resemble describes as imperceptible and EU AI Act compliant. Organizations that cannot send data outside their environment can choose containerized on-premises or air-gapped deployment with zero retention. The product is aimed at teams checking media and meeting participants, rather than general-purpose content creation. Its Flex plan has no subscription fee, but detection is billed by usage.

The main advantage is breadth: Resemble covers audio, images, and video, with meeting detection, identity matching, and watermarking alongside its core detection model. That gives security teams a single platform to evaluate for more than one workflow, rather than stitching together separate tools. Teams can start through the API or SDK, or use the Microsoft Teams integration without adding a meeting bot.

Deployment choice is another practical strength. Resemble offers containerized on-premises and air-gapped options for organizations that need media to stay inside their environment. Enterprise also lists custom model training, volume pricing, SLAs, and dedicated support.

The pricing page lists a low-commitment Flex option: no subscription, one team seat, and pay-as-you-go credits for detection and platform/API access. Team includes five seats, and Business includes 20 plus SSO. Annual billing lowers Team from $350 to $280 per month and Business from $1,000 to $800 per month, a stated 20% saving. Those published prices make it possible to compare the subscription cost before committing.

The $0 monthly price on Flex does not mean detection is free: it has no subscription fee, but charges per use. The pricing page lists audio detection at $0.035 per second on Flex versus $0.015 on Team or Business; video costs $0.07 per second on Flex and $0.03 on the paid plans. Image detection is $0.035 per image on Flex and $0.015 on Team or Business. High-volume video or audio review could therefore make usage costs the bigger line item.

Seat counts are explicit caps on what each plan includes: one on Flex, five on Team, and 20 on Business. The published pricing details do not state included credit allowances, request limits, or file-size limits, so buyers should confirm those before estimating total cost. They also do not publish detection accuracy or false-positive rates. Enterprise pricing is custom, which makes it harder to compare without a sales quote. The flexible deployment story is useful, but the public pricing page leaves important budgeting and performance questions unanswered.

Flex has no subscription fee, but it is not free usage: detection and other services are billed per use. The pricing page lists Flex at $0 per month with one seat. Team costs $350 monthly or $280 per month with annual billing, and includes five seats. Business jumps to $1,000 monthly or $800 annually, with 20 seats, SSO and org-wide calendar integration. Annual billing saves $840 a year on Team and $2,400 on Business.

Usage rates vary by plan and task. Audio detection is $0.035 per second on Flex versus $0.015 on Team or Business; video is $0.07 versus $0.03 per second. Image detection is $0.035 per image on Flex and $0.015 on paid plans. Intelligence for meetings costs $0.025 per second on Flex or $0.015 on Team and Business. Identity search and watermarking are also charged per call. Enterprise pricing is custom. Budget for usage separately: the listed subscription price is not an all-in cost.

Switching effort depends on how you use Resemble. The API is described as a single-call detection flow that returns a deterministic score, so developers replacing another API should map its inputs and outputs to their existing workflow. Teams using the native Microsoft Teams integration or an on-prem deployment will also need to account for those specific setups. Enterprise deployment and custom model training may add project work; the source does not specify migration services or timelines.

Data export, account offboarding and portability of enrolled identity data are not published on the vendor's site. Ask Resemble what can be exported and in what format before moving sensitive workflows. The pricing and product information here does not include Reddit posts, so we can't responsibly say what users report as reasons for leaving. For a switch decision, check usage costs against your actual volume and confirm export terms, retention and deployment requirements with the vendor.