JA EN
shiichan

Pollo AI Turns Ideas into Video Ads with GPT-6 Astra and GPT-Image-2.5!

OpenAI News ★ View Markdown
#openai#ai#business

Hi, it's Shii-chan! Today I'm sharing a video-making story from an OpenAI customer case study!

OpenAI News openai.com

What was announced?

OpenAI News published a case study about Pollo AI.

Pollo AI is a platform built so that people can turn ideas into images and videos without production expertise. More than 26 million users make videos, ads, and visual content with it.

Its latest product, Pollo Agent, uses OpenAI models such as GPT-6 Astra and GPT-Image-2.5 to power a guided video creation workflow.

Why it matters

CEO Bill Zhu wants to do for video what Canva did for design.

Video creation tends to eat time in choosing models and trial and error. Pollo reports that users spend over 50% less time choosing or switching models.

What changes

Pollo users can go from rough ideas and visual references to storylines, scene flows, and scripts.

Here are the usage examples from the story:

  • Photo to Video Ads (Marketing Studio): turns a single photo and a prompt into a short video, for small businesses and DTC brands
  • A jewelry brand: used Pollo Agent and GPT-Image-2.5 to create an editorial campaign image of a model beside a cheetah, at lower time and cost than a live shoot
  • A beverage brand: transformed a peach soda bottle photo into a playful snowy miniature landscape

There's also an example of a fragrance product photo becoming a roughly 15-second luxury ad.

Dive Deep

The model split is described concretely:

  • Routing: handled by GPT-5.6
  • Harder tasks (such as narrative development and scene revisions): handled by GPT-6 Astra, matching reasoning depth to each task to balance quality, speed, and cost
  • Image generation: GPT-Image-2.5 handles posters, storyboards, and campaign visuals

CTO Peter Zhou said OpenAI models stood out for handling ambiguous creative inputs and for their multimodal capabilities. The stability of the API and the release cadence are also valued.

Pollo also tests generative models against real user prompts and workflows, scoring output quality, consistency, stylistic range, speed, and cost.

The main users are content creators, and the platform is popular for micro dramas that mix AI and human actors. About 30% of creator sessions start from templates or trending-format workflows.

Wrap-up

  • Pollo AI is a video and image creation platform with over 26 million users
  • Pollo Agent uses GPT-5.6 (routing), GPT-6 Astra (harder tasks), and GPT-Image-2.5 (images)
  • Time spent choosing or switching models is reported down by over 50%
  • Over the next 12 months, Pollo aims to make AI video part of everyday content creation

This story is for small businesses who want ad videos fast, and for anyone curious about multi-model design!

Related posts