by Deploy (formerly Hive)
Content moderation APIs from Deploy, the company formerly called Hive, scoring images, video, text and audio from a portfolio of 20+ in-house models.
Pick Deploy Content Safety if you run a large social, streaming or dating platform and want one vendor for every content type, including CSAM. Skip it for a small community: self-serve keys stop at 100 requests a day and most models are sales-priced. Same Hive models, new name since late 2026.
Every content type, including audio, OCR and Thorn-backed CSAM detection, under one API account and one review tool.
Deploy says hundreds of large organisations send it billions of pieces of content a day and that it runs its own datacentres (rename blog post, 2026), which keeps big platforms on it, while the move into general model hosting gives moderation specialists an opening.
Main trade-off
Real-volume pricing for anything beyond images and text is hidden behind sales.
| Plan | Monthly price | What it includes |
|---|---|---|
| Pay as you go (self-serve) | Custom | Visual Moderation $3.00 per 1,000 requests, Text Moderation $0.50 per 1,000 requests, 100 requests a day per model, API keys from the Deploy platform |
| Enterprise (contact sales) | Custom | Higher request limits, Video frame sampling for Visual Moderation, Audio and OCR Moderation, CSAM Detection and CSE text classifier, Visual and audio moderation with GARM brand-safety labels, SAML SSO, RBAC and configurable data retention |
Pay as you go, billed per request: Visual Moderation $3.00 per 1,000 requests and Text Moderation $0.50 per 1,000 requests, each capped at 100 requests a day on a self-serve key. Higher limits, video, Audio Moderation, OCR Moderation, CSAM Detection and the CSE text classifier are priced by sales. No monthly plan and no free tier are listed.
Deploy Content Safety is the content moderation API suite formerly sold as Hive Moderation. It scores images, sampled video frames, text, audio and text inside images against policy classes, and adds a Thorn-backed CSAM detector. It draws on a portfolio of more than 20 proprietary models trained by the company.
Visual Moderation
Scores images, and video frames sampled at a rate you set, for nudity, violence, drugs, weapons, hate symbols and other classes, each with its own confidence score.
Text Moderation
Classifies posts, comments and chat for harassment, hate, sexual content, violence and spam, with multi-level severity results instead of a single yes or no.
Audio and OCR Moderation
Transcribes speech and text inside images or video and moderates it, so a slur in a voice clip or on a meme is caught like a typed one.
CSAM Detection and CSE text classifier
Hash matching and classifiers built on Thorn's technology flag known and new child sexual abuse material and grooming or sextortion language for reporting.
Arbiter review tool
The renamed Moderation Dashboard turns model scores into policy rules, queues borderline items for human review and records the action taken.
DeployVLM
A multimodal language model Deploy offers for low-latency moderation, detection and verification, billed per token at $0.50 per million input tokens.
The nearest AI tools on Hokai for the same job.
| Tool | Price from | Free plan | Runs on | Rating |
|---|---|---|---|---|
| Quote | No | Web | 3.6 | |
| Quote | No | Web | 3.9 | |
| $169/mo | Yes | Web | 3.9 | |
| Quote | No | Web | 3.9 |
Proprietary
A self-serve key bills image checks at $3.00 and text checks at $0.50 per thousand calls, with a daily limit of 100 calls per model. Everything else, including higher limits, video, audio, OCR and the child-safety models, is quoted by sales. Video is billed per sampled frame, so the sampling rate you choose drives the cost.
No free tier is listed on the Deploy pricing page. You can try the public visual moderation demo, then pay per request on a self-serve key that is capped at 100 requests a day per model.
Yes. The company announced the rename in late 2026 and widened into hosting open-source language models. Existing Hive accounts and logins remain active, and the Hive Portal became the Deploy platform. The content safety models continue under the Deploy name.
Yes. Its CSAM Detection API and a CSE text classifier are built on technology from Thorn, the child-safety nonprofit. Both are sold through sales rather than self-serve, which suits the vetting that CSAM work requires.
It was renamed Arbiter when Hive became Deploy. Arbiter is the policy enforcement and review tool that turns model scores into actions and routes borderline content to human moderators.
Deploy Content Safety is the moderation suite of Deploy, the San Francisco company that was called Hive until it renamed itself in its blog post "Hive is now Deploy" in late 2026. Founded by Kevin Guo and Dmitriy Karpman, Hive built its business on content moderation models and says it has raised more than $200M and employs over 250 people. Reported customers have included Reddit, Giphy and BeReal. Existing Hive accounts and logins carry over, and the old Hive Portal is now the Deploy platform.
The suite is a set of hosted classifiers you call by API: Visual Moderation for images and sampled video frames, Text Moderation, Audio Moderation, OCR Moderation for text inside images, and a CSAM Detection API plus a child-exploitation text classifier built on technology from Thorn, the nonprofit behind Safer. Each returns per-class scores, so your own rules decide what is removed, blurred or sent to a person. When Hokai opened the public visual moderation demo on 22 November 2026, with no account, its preloaded sample image came back with 71 class scores, led by general_suggestive and yes_male_shirtless at 1.00 and yes_middle_finger at 0.02. Arbiter, formerly the Moderation Dashboard, is the review and policy tool on top, and the same account reaches Deploy's AI-generated content detectors, which sit closer to the AI content detectors shelf.
It fits trust and safety engineers at social apps, livestreaming services, dating apps and marketplaces that need one vendor for every content type at high volume. Teams that run their review queues in a case-management tool such as Cinder can feed Deploy scores into it.
Small communities and side projects should look elsewhere. Self-serve access is rate-limited per day, and video, audio, OCR and CSAM all need a sales conversation. A team that only screens text and the odd image can start with the free endpoint from OpenAI or a developer API with a published monthly plan, such as Sightengine, and a platform that needs EU Digital Services Act reporting built in may prefer a compliance-first platform such as Checkstep. Hokai keeps the wider category in AI governance tools and covers adjacent security buying in the AI governance and security guide.
The rename also widened the company: Deploy now sells hosted open-source language models beside its own classifiers, and says content safety remains a core part of the product suite. Buyers signing a multi-year moderation contract should ask how the moderation roadmap is staffed now that the company sells general inference too.
Product facts on this page (pricing, plans, integrations, compliance) are read from Deploy (formerly Hive)'s own pages and documentation. The verdict is Hokai's reading of them, not the vendor's claim.