Price terms as published

Each line is the vendor’s own wording, split from the price capture checked on 2026-10-22. Nothing here is rounded, converted or re-worded — the full sentences stay in the review below. Lines that carry only a price continue the plan named on the line above, in the vendor’s own order.

4 price terms for Jockey by TwelveLabs
#Plan / termChecked
1Free $0/3 GB2026-10-22
2Plus $29/mo 50 GB2026-10-22
3Pro $99/mo 200 GB2026-10-22
4Enterprise custom2026-10-22

One-liner: Jockey is a video intelligence agent from TwelveLabs that lets you talk to your entire video and photo library — search any moment by meaning, understand what happens inside clips, auto-organize footage, and assemble finished cutline-ready media, all in plain language inside Claude, ChatGPT, or your own product via API/SDK. It went public on 2026-10-15, and pricing (re-read from twelvelabs.io 2026-10-22) runs through TwelveLabs’ standard API plans: Free $0/3 GB, Plus $29/mo 50 GB, Pro $99/mo 200 GB, Enterprise custom.

What Jockey is

Jockey is TwelveLabs’ “agent layer for video understanding.” Where the company’s Marengo and Pegasus models are the raw engines — seeing and understanding video the way humans do — Jockey is the orchestration layer that connects them into one conversational workflow. The official launch release frames it as “video search and reasoning capabilities together in one unified workflow.”

The product page groups its capabilities into four:

  • Search — find any moment by meaning across every video and photo in your library: phrase, image, clip, person, place, action. “No tags, no filenames, just what you’re looking for.”
  • Understand — Jockey reads each clip and explains what is there, scene by scene: scenes, context, captions, topics, sentiment, key moments.
  • Organize — footage sorts itself into meaningful sets (person, place, topic, event, character, project) with no manual foldering or naming conventions.
  • Assemble — ask, and Jockey assembles finished media straight from source footage: clips, collections, timelines, sequences, roadmaps, highlights — refine, trim, and export as MP4.

It is positioned for two audiences. Creators and editors get “your whole archive, on tap”: describe a moment and find the exact clip, find any quote you have ever said on camera, or get an edit-ready highlight reel by topic. Marketing teams get ad intelligence aimed at a creative library: automatic ad tagging by hook, talent, format, and tone; winning-hook detection on the opening three seconds; talent and logo recognition; format and trend breakdowns; social and trend intelligence from a current YouTube/TikTok/Instagram Reels corpus; performance insights; and concept-video previews that turn a campaign direction into a playable 15-30s spot.

Pricing (read from twelvelabs.io/pricing, 2026-10-22)

Jockey does not have its own price list — it is delivered through TwelveLabs’ API platform plans, which include storage, media processing, and access to TwelveLabs APIs (Marengo, Pegasus, and the Jockey agent work).

Plan Price (monthly) Features
Free $0/month 3 GB cumulative storage; Basic workload; basic rate limit; shared environment; usage capped at plan limits
Plus $29/month 50 GB cumulative storage; Medium workload; high rate limit; shared environment
Pro $99/month 200 GB cumulative storage; High workload; high rate limit; optional overage
Enterprise Custom Custom storage and workload; overage per contract; SSO/SAML included

Annual billing shows a 17% discount on the Plus and Pro prices. All larger plans describe usage as capped at plan limits, with Pro adding an optional overage — the volume-metered structure of an API product rather than a per-seat SaaS.

Launch timeline and the agentic shift

Jockey’s public launch on 2026-10-15 closed a research preview that began in July: TwelveLabs’ own Product Hunt page lists “Jockey by TwelveLabs — The video AI agent that understands your whole library” as launched July 21, 2026, and the YouTube introduction is dated 2026-07-20 (“Jockey is now in research preview”). The October press release calls this the official launch that moves video intelligence into the broader AI workflow. The company’s engineering blog (“Video Intelligence is Going Agentic”) explains the underlying bet: in 2026, machines can reason across an entire video and photo library by person, by moment, by place — the way people actually think about their libraries — with Jockey combining video and photo foundation models with LLM-based reasoning through a purpose-built agent architecture.

A useful distinction: Jockey is not a standalone consumer app with its own login wall. It “works in Claude, ChatGPT” — usable through those assistants (via MCP per the video intro) — and developers can plug it into their own products through the Jockey API or SDK, with the ability to ingest YouTube, TikTok, and Instagram Reels links and create shareable video-driven artifacts with playable clips.

Independent cross-check (2026-10-22)

  • The official launch date is confirmed by the press release distributed via GlobeNewswire on Oct 15, 2026: “TwelveLabs, the leading video intelligence company, today announced the official launch of Jockey by TwelveLabs.” Note the press release is paid vendor content — it confirms the launch, not an independent evaluation. (third-party distribution, vendor-authored)
  • Product Hunt lists Jockey’s research-preview launch on July 21, 2026, describing it as “the first AI that understands your entire media library across photos and video,” usable in Claude or ChatGPT or via API/SDK, with YouTube/TikTok/Instagram Reels ingestion and MP4 assembly. (third-party)
  • The company’s own product page and blog agree on the four capabilities and the “agent layer” positioning; the blog’s “going agentic” essay was published alongside the public launch. (vendor)
  • The pricing page is titled “API Plans and Costs” and shows the same table for both monthly and annual views — consistent with TwelveLabs’ platform-model positioning. (vendor)

What we did not test

We have not uploaded media to a TwelveLabs account, run a real search against a personal library, or measured search/assembly quality on actual footage. This is an official-information entry with no score — every number comes from the vendor’s pricing and product pages as fetched 2026-10-22, plus the third-party launch coverage listed. The open questions a real trial would answer: how accurately “search by meaning” handles ambiguous descriptions, whether ad-hook detection generalizes beyond the showcased verticals, and whether the 3-50-200 GB storage ladder fits a working creator’s monthly footage volume.

More AI media reading on this site

  • Rill Browser — an agent workspace, not a media tool, but the “agent over a big corpus” pattern closest to Jockey. (review in progress)
  • Spira Maxima — a social-video tool whose pricing page we track quarterly; Jockey is the search-understand side of video, Spira Maxima the generate-publish side.
  • ReelMimic — local multi-agent video style cloning, the self-hosted end of the video-AI spectrum.

Sources