Breaking

GPT-6 Astra Demonstration Shows a 14-Year Editor Automating Routine Video Cuts

September 11, 2026 at 13:16 EDT

AI Agents
Media Generation
Foundation Models

A video editor with 14 years of experience has demonstrated GPT-6 Astra carrying out a multi-step editing workflow, including reviewing source clips, trimming material, selecting music and generating subtitles in DaVinci Resolve. The reported workflow highlights Astra's computer-use capabilities while also exposing the gap between fast assembly and editorial judgment.

SEPTEMBER 2026 · OPENAI

GPT-6 Astra brings routine video editing into an agent workflow

A reported workflow had Astra inspect 84 clips, make edits, find music and generate subtitles in DaVinci Resolve. The speed gains are real, but creative judgment still matters.

clips in the reported workflow

72.6%

OSWorld 2.0 score

47%

less OSWorld time than GPT-5.6 Sol

1.05M

token context window

A measured speed edge

Mind2Web task completion, relative to GPT-5.6 Sol.

1.0x

GPT-5.6 Sol

1.9x

GPT-6 Astra

Faster computer-use completion does not guarantee finished editorial quality.

How the workflow fits together

Inspect footage

→

Select and trim

→

Add music and captions

→

Human review

Where it helps

Clip triage, transcription, repetitive edits and a faster first rough cut when the direction is clear.

Where it stops

Pacing, narrative coherence, color consistency and final creative decisions still need an experienced editor.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.