Video Editing Agent invideo Triples Color-Grading Success with GPT‑6 Astra!

Hi, it's me, Shiichan! Today I've got an exciting video-editing story from OpenAI News.

OpenAI News openai.com

What was announced?

OpenAI News published a customer story about invideo, an agentic video editor built to handle work like shaping the story, placing sound effects, building transitions, and adjusting color, while keeping the human editor in control. With GPT‑6 Astra, invideo now plans and executes complex edits with much greater precision.

Here's what CEO Sanket Shah had to say:

"How Astra can plan a particular edit on a frame-level accuracy is quite stunning."

The story so far

For an AI agent to handle editing, it has to translate the editor's direction into a sequence of steps, pick the right tools, execute the work, and verify the result. Color work is especially tricky, since the agent has to choose from overlapping approaches like correction, grading, regeneration, LUTs, and isolation.

According to Sanket, invideo previously saw "very high failure rates for color-grading and color-correction tasks." Tasks like changing a background while preserving a person's skin tone require isolating and tracking the person across frames before applying color elsewhere, which was hard to get right.

What changes

With GPT‑6 Astra, invideo's success rate on color-grading and color-correction tasks improved about three times. The model also plans complex operations using fewer reasoning steps and output tokens, and it can carry out several instructions in a row without losing the editor's original objective as tasks grow longer and more complex.

Dive Deep

The article highlights three strengths of GPT‑6 Astra:

• Planning complex edits: it breaks the editor's direction into steps and places changes at the correct points in the timeline, using fewer reasoning steps than models invideo previously tested
• Choosing the right approach for color work: it picks from correction, grading, regeneration, LUTs, and isolation depending on the situation, including tracking a person across frames while changing only the background
• Creating custom effects: it turns descriptions and visual references into effects coded specifically for the footage, places them on the timeline, and adds controls so editors can refine them further. In one day, a few invideo editors used it to create about 50 effects

With this setup, editors keep control over the story, their taste, and the final cut, without losing hours to tedious planning and frame-by-frame work.

Wrap-up

• OpenAI News published a customer story about invideo adopting GPT‑6 Astra for its agentic video editor
• Color-grading and color-correction success rates improved roughly threefold
• The model plans complex edits in fewer reasoning steps and stays on track even as tasks grow longer
• A few invideo editors created about 50 custom effects in a single day
• Worth a look for creators wanting to streamline video editing, and anyone curious about real-world agentic tool use cases