Reality Defender vs Sensity: deepfake detection on the same 14 fields and the deepfake fields
Reality Defender publishes a price and Sensity does not: Reality Defender's API page lists a Free plan of 50 image scans a month and a Business plan at $399 on annual billing for 1,000 scans a month of image, audio and video, while Sensity's product page offers a trial button and a sales form and no price, quota or trial term, by the vendor pages read 9 November 2026 (realitydefender.com/api, sensity.ai). Neither card has a study record: none of the 35 records in the studies index scored either product, and the three challenge records on voice and video in the index (ASVspoof 5, SAFE 2025, DFDC 2020) name anonymised teams, not these vendors. The two cards are read side by side below on the same 14 fields and then on the six fields a deepfake buyer asks about first: modalities, file types, retention, deployment, compliance and whether a price is published.
Updated Fields checked 9 November 2026
Short answer
Both detect video, image and audio by API; Reality Defender adds a text line (.txt up to 900 KB) and Sensity adds file forensics (metadata, provenance and container structure) and three on-premise forms, a server, a single-investigator workstation and a USB dongle for air-gapped environments (realitydefender.com/faq, docs.sensity.ai, read 9 November 2026). Reality Defender names 15 upload formats and Sensity 22, the most in the index; Reality Defender publishes a Free plan and a $399 Business plan and Sensity publishes no price; neither states a retention period for submitted media, neither names a security standard, and neither publishes an accuracy or false positive rate (docs.realitydefender.com, docs.sensity.ai/resources/formats-limits, read 9 November 2026). No indexed study scored either, so every figure on this page is the vendor's own. No detector score should be the only evidence in a misconduct, hiring or publishing decision.
How do Reality Defender and Sensity compare on the 14 fields?
| Field | Reality Defender | Sensity |
|---|---|---|
| 1. Name, vendor, official domain | Reality Defender deepfake detection, Reality Defender, Inc., realitydefender.com | Sensity AI Deepfake Detection, Sensity B.V., sensity.ai |
| 2. Modalities detected | video, image, voice (audio), text | video, image, voice (audio), file forensics (metadata, provenance and container structure) |
| 3. Generators covered | Not named by generator; manipulation types listed: face swaps, generated video, altered faces, synthetic speech on the audio track, AI-agent callers | Face manipulation (face swap, lip sync, reenactment, morphing); AI-generated content (AI avatars, synthetic GAN faces, text-to-image, text-to-video); synthetic voice and voice cloning; a Model Generator attribution shown in an example (Flux); no generator list |
| 4. Languages supported | ||
| 5. Input limits | SDK local files: images .jpg, .jpeg, .png, .gif, .webp up to 50 MB; audio .mp3, .wav, .m4a, .aac, .ogg, .flac, .alac up to 20 MB; video .mp4, .mov up to 250 MB and 30 minutes; text .txt up to 900 KB; social media URLs from six platforms; the Free plan supports image uploads only | Files up to 500 MB by upload or public URL; audio and video up to 1 hour; video up to 2560 px; 7 image, 8 video and 7 audio formats; URLs from eleven platforms; 100 requests a minute with a workspace quota per billing period; on-premise deployments enforce no fixed media limits |
| 6. Free tier | Free API plan: $0 a month, 50 scans a month, up to 3 seats, image analysis only, API keys and an analytics dashboard | None published; new cloud accounts are held for manual review and cannot log in or create API keys until approved |
| 7. Paid plans | Business $399 on annual billing as shown: 1,000 scans a month, 1 seat, image, audio and video; Enterprise custom: custom volume, unlimited seats, dedicated support; RealCall and RealMeeting through sales | Cloud platform and API by sales, quota per billing period (monthly, quarterly or yearly); on-premise server, workstation or USB dongle by sales; no price published |
| 8. API | Yes, docs.realitydefender.com | Yes, docs.sensity.ai |
| 9. Integrations | SDKs for Python, TypeScript, Go, Rust and Java; REST API with webhooks; RealMeeting for Microsoft Teams; RealCall in IVR and live contact-centre calls; RealScan browser review; social media URL submission | REST API v2 with signed upload URLs, webhooks and CSV export; SDK; web app with teams, roles and analytics; URL submission from eleven platforms; on-premise server, workstation and USB dongle; integration at evidence ingestion in forensic environments |
| 10. Vendor accuracy and false positive claims | No numeric figure; "an inference system that grades each content with a 1-99% probability rating"; a Gartner designation of 25 June 2026 and a bank case study of 2.4M+ calls with about 1,010 synthetic voice calls detected are cited | No numeric figure; "The Most Accurate Deepfake Detection Solution", "Unparalleled accuracy", "lower false positives"; the 89.0% and 99.0% on the tech page are confidence scores in example panels |
| 11. Independent studies | 0 records | 0 records |
| 12. Data retention and privacy statement | Privacy policy (10 August 2026): personal information kept no "longer than is reasonably necessary", no fixed period; photos, videos, audio files and voice recordings are among the information collected; the FAQ states "a range of US, UK and EU certifications" without naming one | Privacy policy (January 2024): data kept "for as long as you use our Services or as necessary"; biometric data (selfie videos and identification documents) deleted within three years of last interaction; no period for submitted media; submissions deletable through the API |
| 13. Launch year and ownership | 2021; Reality Defender, Inc., grown out of a nonprofit research initiative; no parent stated | 2018; Sensity B.V., Amsterdam, co-founders Giorgio Patrini and Francesco Cavalli; no parent stated |
| 14. Rankings the card appears in | Studies by content type (0 records), platform integrations, file types | Studies by content type (0 records), platform integrations, file types (first, 22 formats) |
Which studies tested Reality Defender or Sensity?
None. No record in the studies index ran either product, and the index enters no vendor figure as a study result. The three records on voice and video in the index are challenges: ASVspoof 5 (53 anonymised teams on 542,086 spoofed utterances), SAFE 2025 (12 teams on 13 text-to-speech models) and DFDC 2020 (2,114 Kaggle teams on 10,000 black-box videos), and none names Reality Defender or Sensity as a participant (arXiv 2601.03944, arXiv 2510.03387, ai.meta.com, read 9 November 2026). The image records, Ha and Passananti 2024 and RAID image 2025, scored AI or Not, Hive, Illuminarty and Sightengine, not these two. Both cards therefore sit at 0 records in studies by content type, and the deepfake fields below are the comparison.
| Deepfake field | Reality Defender | Sensity |
|---|---|---|
| Modalities | Video, image, voice (audio), text (.txt up to 900 KB) | Video, image, voice (audio), file forensics (metadata, provenance, container structure) |
| File types stated | 15: jpg, jpeg, png, gif, webp; mp3, wav, m4a, aac, ogg, flac, alac; mp4, mov; txt | 22: jpg, jpeg, png, tiff, gif, webp, jfif; mkv, flv, mov, mp4, avi, webm, 3gpp, m4v; wav, mp3, m4a, ogg, aac, flac, pcm |
| Size and duration ceilings | Images 50 MB; audio 20 MB; video 250 MB and 30 minutes; text 900 KB | Files 500 MB; audio and video 1 hour; video 2560 px on the longest side; no fixed limits on-premise |
| Retention of submitted media | No fixed period; "not retain your personal information for longer than is reasonably necessary" | No period for submitted media; biometric data deleted within three years of last interaction; submissions deletable by API |
| Deployment | Cloud API and SDKs; RealMeeting inside Microsoft Teams; RealCall inside contact-centre telephony; no on-premise statement on the pages read | Cloud API and web app; on-premise server on dedicated NVIDIA GPUs; on-premise workstation; USB dongle for field teams and air-gapped environments |
| Compliance standards named | None; the FAQ states "a range of US, UK and EU certifications" without naming one | None named on any page read |
| Pricing published | Yes: Free $0 (50 image scans a month), Business $399 on annual billing (1,000 scans a month), Enterprise custom | No: trial and sales forms only; accounts approved manually |
| Verdict format | A 1 to 99% probability rating per item, "without the need for the ground truth" | A binary verdict, suspicious or valid, aggregated conservatively so that any suspicious detection makes the whole submission suspicious, with confidence scores and heatmaps |
Fields as recorded on the two cards from the vendor pages named above, read 9 November 2026; the file types ranking sorts the format counts, the voice and video question lists the other cards that state both media, and the compliance view records that neither vendor names a standard.
What do the two cost over 12 months?
| Plan | Monthly as stated | Annual as stated | Inclusions as stated | Checked on |
|---|---|---|---|---|
| Reality Defender Free | $0 | 50 scans a month, up to 3 seats, image analysis only, API keys, analytics dashboard; "for trialing the API" | 9 November 2026 | |
| Reality Defender Business | $399, annual billing as shown | 1,000 scans a month, 1 seat, image, audio and video analysis, results plus supporting detail, analytics dashboard, unlimited API keys and SaaS web platform, self-serve onboarding and chat support | 9 November 2026 | |
| Reality Defender Enterprise | Custom | Custom | Custom scan volume, unlimited seats, image, audio and video, unlimited API keys, dedicated support; RealCall and RealMeeting through sales | 9 November 2026 |
| Sensity cloud platform and API | No public price | No public price | Web app with teams and roles, REST API v2, SDK, webhooks, CSV export; submission quota per billing period (monthly, quarterly or yearly) | 9 November 2026 |
| Sensity on-premise | No public price | No public price | Server on dedicated NVIDIA GPUs, single-investigator workstation, or USB dongle for field teams and air-gapped environments | 9 November 2026 |
Prices in USD as shown on realitydefender.com/api, checked 9 November 2026; the Business figure is printed beside "Annual Billing" and the page does not state a monthly equivalent, so the index does not compute one. Sensity publishes no price on any page read (sensity.ai, docs.sensity.ai, read 9 November 2026), and its documentation states that new cloud accounts wait for approval by a Sensity administrator before an API key can be created.
One of the comparisons in the compare hub. The cards are Reality Defender and Sensity, and the home board answers which AI detectors exist and who has tested them? A fraud, trust-and-safety or forensic team weighing two deepfake detection vendors that both sell by API and to institutions is the reader this pair is written for; the voice and video question lists every card that states both media, and the government view reads deployment and data region for the public-sector buyer.
Which card fits which reader?
- Choose Reality Defender ifyou want a published price and a self-serve start: a Free plan of 50 image scans a month and a $399 Business plan on annual billing (realitydefender.com/api, read 9 November 2026). Sensity publishes no price and approves accounts by hand.
- Choose Sensity ifthe media cannot leave the building: an on-premise server, a workstation and a USB dongle for air-gapped environments are stated, with no fixed media limits on-premise (docs.sensity.ai, read 9 November 2026). Reality Defender's pages state cloud, Teams and telephony deployments and no on-premise option.
- Choose Sensity ifthe evidence arrives in many containers: 22 formats including tiff, jfif, mkv, flv, avi, webm, 3gpp, m4v and pcm, against Reality Defender's 15 (docs.sensity.ai/resources/formats-limits, docs.realitydefender.com, read 9 November 2026).
- Choose Reality Defender ifthe threat is a live call or meeting: RealCall runs in IVR and live contact-centre calls and RealMeeting inside Microsoft Teams (realitydefender.com/api, read 9 November 2026). Sensity states file and URL submission.
- Choose Sensity ifyou want file forensics beside the detector: metadata, provenance and container analysis run on every submission as a fourth service (docs.sensity.ai, read 9 November 2026).
- Choose Reality Defender ifa probability per item suits your review queue, and Sensity if a single conservative suspicious-or-valid verdict with heatmaps does: the two formats are stated on each vendor's pages and neither is tested in the index.
- IfChoose neither on accuracy: no indexed study scored either product, neither publishes a false positive rate, and the index records no vendor accuracy figure for either card.
How does the evidence compare, tool by tool?
Reality Defender
- Register ID
- AIDI-D-0022
- Study records
- 0 in the index
- Shared with Sensity
- 0
- Vendor accuracy claim
- We use an inference system that grades each content with a 1-99% probability rating -- all without the need for the ground truth. (No numeric accuracy or false positive figure published; the home page cites a Gartner Emerging Market Quadrant for Deepfake Detection Market Shaper designation of 25 June 2026 and a bank case study of 2.4M+ calls analyzed in 2025 with about 1,010 synthetic voice calls detected.)
- Free tier
- Free API plan: $0/month, 50 scans per month, up to 3 seats, image analysis only, API keys and an analytics dashboard; audio and video need a paid plan
- Languages
- Not stated
Sensity
- Register ID
- AIDI-D-0038
- Study records
- 0 in the index
- Shared with Reality Defender
- 0
- Vendor accuracy claim
- No accuracy percentage is published. Home page: The Most Accurate Deepfake Detection Solution; title tag: Best Deepfake Detection Software in 2026; product page: Unparalleled accuracy in deepfake detection; By combining multiple detection techniques, we ensure higher accuracy, lower false positives, and robust adaptability against evolving threats. The 89.0% and 99.0% figures on the tech page are confidence scores in example result panels, not accuracy rates. Counters on the home page: 35 min saved on manual reviews per case and +30 countries where we are deployed.
- Free tier
- None published. The site offers Get started and Start a trial buttons and a Talk to sales form; new cloud accounts are held for manual review and cannot log in or create API keys until approved (documentation). No prices, quotas or trial terms are published
- Languages
- Not stated
How this comparison is built
- Cards compared
- Reality Defender and Sensity
- Fields
- The 14 fields of every card, in card order
- Shared study records
- 0
- Fields checked
- 9 November 2026
- Records in the studies index
- 35 records
Both columns are drawn from the same 14 fields in the index data file, read on 9 November 2026 from the vendor page recorded on each card, and from the studies index, which holds no record for either product; the second table repeats the six fields a deepfake buyer asks about first, each copied from the card field that carries it. Where a vendor publishes no figure, the cell says so rather than carrying a number from a third party: Sensity's price, both vendors' accuracy, both retention periods for media and both compliance lists are absent on the pages read, and the comparison states each absence once. The rules, the recheck schedule (next pass 8 December 2026) and the correction route are on the methodology page; the independently tested ranking lists the cards that do have a record, and the false positive rate ranking explains why a vendor's "lower false positives" is not a rate.
The index's second media-only pair is the Pindrop vs Resemble Detect comparison, two voice vendors with no study record between them, read 10 November 2026.
Where each figure comes from
Values as each vendor states them on realitydefender.com, its FAQ, API page, SDK quickstart and privacy policy, and on sensity.ai, docs.sensity.ai, its formats and limits page and privacy policy, all read 9 November 2026. An empty cell means the card has no published value for the field: neither vendor states a language count, because neither detects text in the way a text detector does (Reality Defender's .txt line is stated as a file type, not a language list).
Questions about Reality Defender and Sensity
Has any study tested Reality Defender or Sensity?
No indexed study has scored either product; both cards sit at 0 records in studies by content type. The voice and video records in the index are challenges with anonymised teams (ASVspoof 5, SAFE 2025, DFDC 2020), and the image records scored other vendors.
Which publishes a price?
Reality Defender: Free at $0 for 50 image scans a month and Business at $399 on annual billing for 1,000 scans a month, with Enterprise by quote (realitydefender.com/api, read 9 November 2026). Sensity publishes no price on any page read.
Which runs on-premise?
Sensity states an on-premise server, a workstation and a USB dongle for air-gapped environments (docs.sensity.ai, read 9 November 2026). Reality Defender's pages state cloud API and SDK deployment, Teams and telephony integrations, and no on-premise option.
Which accepts more file types?
Sensity, 22 against 15, by the formats each vendor lists (docs.sensity.ai/resources/formats-limits, docs.realitydefender.com/sdks/quickstart, read 9 November 2026); Sensity's 22 is the largest count in the file types ranking.
Does either state how long submitted media is kept?
Neither states a period for media. Reality Defender keeps personal information no "longer than is reasonably necessary" and Sensity's three-year clause governs biometric data from identity checks, not submitted files (realitydefender.com/t-c/privacy-policy, sensity.ai/privacy-policy, read 9 November 2026).
Does either name a security standard?
No. Reality Defender's FAQ states "a range of US, UK and EU certifications" without naming one, and Sensity names none on the pages read, so neither card appears in the compliance view.
Which detects text as well as media?
Reality Defender lists text (.txt up to 900 KB) among its SDK inputs (docs.realitydefender.com/sdks/quickstart, read 9 November 2026); Sensity does not detect text and adds file forensics instead.
Sources and revisions
Sources
- Reality Defender homerealitydefender.com
- Reality Defender FAQrealitydefender.com/faq
- Reality Defender API and plansrealitydefender.com/api
- Reality Defender SDK quickstart (formats and limits)docs.realitydefender.com/sdks/quickstart
- Reality Defender API referencedocs.realitydefender.com/api-reference/qu...
- Reality Defender privacy policyrealitydefender.com/t-c/privacy-policy
- Sensity deepfake detection product pagesensity.ai/deepfake-detection-for-video-i...
- Sensity documentationdocs.sensity.ai
- Sensity formats and limitsdocs.sensity.ai/resources/formats-limits
- Sensity privacy policysensity.ai/privacy-policy
- Sensity termssensity.ai/terms-and-conditions
- ASVspoof 5 evaluation (Wang et al. 2026)arxiv.org/abs/2601.03944
- SAFE challenge (Trapeznikov et al. 2025)arxiv.org/abs/2510.03387
- Meta AI, Deepfake Detection Challenge resultsai.meta.com/blog/deepfake-detection-chall...
