TwelveLabs Launches Pegasus 1.5 and Rodeo at NAB Show — Video AI Gets a Major Platform Upgrade
TwelveLabs unveiled Pegasus 1.5, a new video intelligence model, and Rodeo, a full platform for video analysis and search, at the NAB Show 2026. The releases push multimodal video understanding significantly forward, with improved scene comprehension, temporal reasoning, and production-ready API tooling.
Original sourceTwelveLabs used NAB Show 2026 — the broadcast and media technology industry's biggest annual event — to announce two major launches: Pegasus 1.5, its latest video intelligence model, and Rodeo, a new platform that wraps the model in production-ready infrastructure for media companies and developers.
Pegasus 1.5 represents a substantial jump in video understanding capability over its predecessor. The model can now reason across longer video segments (up to 2 hours in a single context), identify and track objects across non-contiguous scenes, and generate structured metadata — including shot lists, transcripts, and scene descriptions — automatically from raw footage. Temporal reasoning has been a persistent weakness in video AI, and TwelveLabs claims Pegasus 1.5 closes most of the gap with what a human editor would catch.
The Rodeo platform is the more commercially significant announcement. It provides media companies with a turnkey pipeline: ingest raw video, run Pegasus analysis, and expose the results through a search API that understands natural language queries ("find all scenes where a red car appears in rain"). Built-in connectors cover major broadcast archive formats, cloud storage systems, and MAM (media asset management) platforms.
For the broadcast industry, this matters because the archive problem is massive and unsolved. Major networks have decades of video content that's essentially invisible because it's not searchable. Rodeo's pitch is that you can run Pegasus over your archive once, generate structured metadata, and then query it with natural language forever after — transforming a dead archive into a live asset library.
The pricing hasn't been announced publicly, but TwelveLabs has been targeting enterprise deals and has raised over $77M to date. Pegasus 1.5 will be available on the API this summer, with Rodeo launching in private beta for existing customers immediately.
Panel Takes
The Builder
Developer Perspective
“The 2-hour context window for video is the headline feature here — most video AI I've used falls apart on anything longer than a few minutes. If Pegasus 1.5 actually maintains coherence across a full-length film or broadcast segment, this opens up a completely different class of video applications that weren't technically feasible before.”
The Skeptic
Reality Check
“TwelveLabs is well-funded but operating in a space where Google, AWS, and Azure all have competing video AI products with massive infrastructure advantages. The specialized model quality may be better today, but the cloud giants will close the gap. The real question is whether the enterprise sales motion and media-specific features can build enough moat before commoditization arrives.”
The Futurist
Big Picture
“The searchable video archive problem is worth billions in latent value across every media company on earth. Most of that content is currently inaccessible not because it doesn't exist but because it's not indexed. Rodeo is making a bet that whoever wins the video understanding layer owns a permanent toll road on the media industry's past and future content.”