AWS Bedrock adds TwelveLabs' Pegasus 1.5 for structured video search
In short: AWS has rolled out Pegasus 1.5, a video-language model built by TwelveLabs, as a fully managed option inside Amazon Bedrock. Unlike earlier versions that answered one question per clip, this release scans an entire video and outputs a JSON list of every moment matching criteria you define, complete with start and end times. It also handles longer recordings — up to two hours in one call — and lets you feed in a video straight from an S3 bucket or as a base64 string, skipping any indexing step.
This summary was generated automatically by AI from AWS Machine Learning's publication. It is our own text, not a copy of the original — facts, figures and quotes belong to the source, linked above and below.
What changed?
- 1Finds and timestamps every matching moment across a full video, not just answers to a single question
- 2Lets you combine a written description with a reference photo to pinpoint a specific object, logo or person and its on-screen action
- 3Accepts video directly from S3 or inline base64, with no separate indexing pipeline needed beforehand
- 4Raises the per-request video length limit from one hour to two hours compared with Pegasus 1.2
- 5Runs through the InvokeModel and streaming InvokeModelWithResponseStream APIs, currently in the US West (Oregon) region
| Parameter | Before | Now |
|---|---|---|
| Max video length | 1 hour (Pegasus 1.2) | 2 hours per request |
| Output format | Single text answer per clip query | Structured, timestamped JSON segments matching defined criteria |
| Input methods | Pre-indexed clips | Direct S3 URI or base64 string, no pre-indexing |
| Prompting | Text-only queries | Text plus reference image (multimodal) prompting |
| APIs | Not specified | InvokeModel, InvokeModelWithResponseStream (US West/Oregon) |
Why it matters
Teams that need to mine hours of recorded footage — broadcasts, sports, surveillance, training sessions — for specific events can now get structured, searchable metadata back instead of manually scrubbing through video or scripting a Q&A loop per clip.
What it means for AI agents and contact centers
This model works on video rather than phone-call audio, so it doesn't directly apply to call transcription or voice pipelines; it's more relevant if video recordings of meetings or site visits ever need to be indexed for search.
Sources
- AWS Machine LearningOfficialPrimary sourceOriginal article →„Pegasus 1.5 from TwelveLabs is now available in Amazon Bedrock“19 Oct 2026, 19:49
- Published by source
- 19 Oct 2026, 19:49
- Found by our system
- 19 Oct 2026, 20:14
- Summary generated
- 19 Oct 2026, 20:15
This article was written by AI from the original source. Facts, numbers and prices come from the source; missing values are marked “Not specified”. Legal notice, copyright and privacy