TwelveLabs Pegasus 1.5 lands on Amazon Bedrock: turn 2-hour videos into timestamped JSON

AWS ML Blog · rss · 2026-10-20

AWS ML Blog announces TwelveLabs' Pegasus 1.5 video language model is now available on Amazon Bedrock (us-west-2, model ID twelvelabs.pegasus-1-5-v1:0). New in 1.5:

• Time-based metadata generation: define custom segment criteria (editorial segments, sports plays, speaker changes, brand appearances) and get every matching segment back as JSON with start/end timestamps
• Multimodal prompting: combine text descriptions with reference images to locate specific products, logos, or people
• On-the-fly processing: analyze directly from an S3 URI or inline base64 (≤25MB) with no pre-indexing pipeline
• Extended capacity: up to 2-hour videos per request, doubled from 1 hour

The post walks through a full getting-started guide: IAM policies, Boto3 examples for general-mode Q&A and streaming responses, the two analysis modes (general / timebasedmetadata) with field references, checking stopReason for truncated outputs, and tagging application inference profiles to track costs per app.

Original post →

More from Infra

• Together Launches Post-Training Service for Custom Models via RL and SFT — togethercompute · 2026-10-20
• CoreWeave Stock Down 52% in a Year, Back Near IPO Price — bookwormengr · 2026-10-20
• Report: Liang Wenfeng planned independent DeepSeek compute chips before Huawei persuaded him otherwise — teortaxesTex · 2026-10-20
• NVIDIA's Ian Buck explains how AI factories can flex power use to relieve the grid — nvidia · 2026-10-20
• p99 container startup latency cut 63%, saving ~19 hours of compute daily on Cloud Run — DanielLockyer · 2026-10-20
• Late Interaction in 2026: Experiment Shows Single-Vector Search Ranks Wrong Answers Higher — IgorCarron · 2026-10-20

Infra channel →