← Back
AWS ML Blog

Pegasus 1.5 from TwelveLabs is now available in Amazon Bedrock

•16 min read•
#amazon#bedrock#llm
Level:Advanced
For:AI Engineers
✦TL;DR

TwelveLabs’ Pegasus 1.5 video language model is now offered on Amazon Bedrock, enabling on‑the‑fly analysis of videos up to two hours long and returning timestamped segment metadata as JSON. The model adds time‑based metadata generation, multimodal prompting with visual references, and eliminates the need for pre‑indexing by accepting Amazon S3 URIs or base64 strings directly. Engineers can invoke Pegasus 1.5 via the Bedrock runtime InvokeModel or InvokeModelWithResponseStream APIs, keeping data within the AWS boundary.

⚡ Key Takeaways

  • Pegasus 1.5 processes videos up to **2 hours** per request (vs. 1 hour in Pegasus 1.2).
  • Supports **multimodal prompting** where segment definitions combine text and reference images.
  • Returns **structured JSON** with start/end timestamps for every matching segment.
  • Invoked through **Bedrock‑runtime InvokeModel** or **InvokeModelWithResponseStream** APIs.
  • No pre‑indexing required; videos can be supplied via **Amazon S3 URI** or **base64** payload.
💡 Why It Matters

Engineers can now add searchable, timestamped video insights to applications without building custom ingestion pipelines, reducing latency and operational overhead while staying within AWS data residency.

✅ Practical Steps

  1. Call the model on Bedrock using the **InvokeModel** API with `ModelId = "twelvelabs.pegasus-1-5-v1:0"` and provide an S3 URI or base64 video payload.
  2. For streaming results, replace InvokeModel with **InvokeModelWithResponseStream** to receive JSON segments as they are generated.
  3. Define segment criteria using a JSON payload that includes text descriptions and optional reference images to leverage multimodal prompting.

Want the full story? Read the original article.

Read on AWS ML Blog ↗

More like this

Introducing workload recovery and expanded infrastructure checks on Amazon SageMaker HyperPod

AWS ML Blog•#deployment

Introducing Gemini 4 Neon

Google DeepMind•#llm

DaisyRose-27B: Pushing the Frontier of Post-Training Efficiency

Hugging Face Blog•#llm

NVIDIA RTX Spark Compact Desktops Out Now

NVIDIA Blog•#compute

EXPLORE AI NEWS

Daily hand-picked stories on LLMs, RAG, agents and production AI — curated for engineers who ship.

BROWSE NEWS

GET THE WEEKLY DIGEST

Join engineers getting the Monday signal-over-noise AI breakdown. No spam, unsubscribe anytime.

LEARN AI ENGINEERING

Curated courses, research papers, repos and tutorials built for engineers leveling up in AI.

START LEARNING