Invideo is an agentic video editor intended to automate many editing tasks while leaving creative control with the human editor. The company reports that integrating GPT‑6 Astra helps the system plan and execute more complex edits with greater precision, placing changes at the appropriate points on the timeline.
Fewer reasoning steps and steadier objective-following
The AI agent must translate an editor’s directions into a sequence of operations: choose tools, perform actions, and verify outcomes. According to Sanket, GPT‑6 Astra uses fewer reasoning steps and output tokens to complete complex work than models invideo previously tested: “Astra is using far fewer reasoning sets and thus output tokens to complete complex work.”
Invideo also found that Astra is better at maintaining the editor’s original objective as tasks grow longer and include multiple instructions, allowing it to carry out several steps without losing focus.
Threefold improvement in color grading and correction
Astra proved especially effective for color tasks, where the agent must pick among overlapping approaches such as correction, grading, regeneration, applying LUTs, or isolating elements. One common challenge is changing a background while preserving a person’s skin tone — a job that requires isolating and tracking the person across frames before applying color changes elsewhere.
“Earlier, we saw very high failure rates for color‑grading and color‑correction tasks. With Astra, the success rate improved about three times,” Sanket said.
Rapid creation of custom effects
Beyond selecting existing tools, Astra can convert textual descriptions and visual references into custom effects tailored to the footage. The agent can code an effect, place it on the timeline, and add controls so the editor can refine it. In one day, a few invideo editors used the model to create roughly 50 effects.
Why this matters
Integrating GPT‑6 Astra into the invideo agent reduces the time spent on tedious planning and frame‑by‑frame adjustments while preserving editors’ control over story, taste, and the final cut. This can yield substantial time savings and quality improvements in workflows that require precise color handling and bespoke visual effects.
