OpenAI made its new GPT-5.6 model lineup generally available around July 9, 2026. The flagship GPT-5.6 Sol can convert a screen recording directly into a motion design video inside the Codex desktop environment, and it ships alongside 100 ready-to-use video editing skills.
July 9, 2026 · OpenAI
GPT-5.6 Sol Turns a Screen Recording Into a Finished Video
OpenAI's new GPT-5.6 lineup ships generally available across three tiers. The flagship Sol converts a dragged-in screen recording into a motion-design video inside Codex — editing, motion graphics, audio and verification, from natural-language instructions alone.
3
tiers — Sol (performance), Terra (balanced), Luna (cost-efficient)
100
ready-to-use video editing skills: subtitles, voiceover, style changes, translation
500M
tokens reached in ¼ the time of the previous generation
Efficiency vs Claude Fable 5
Same job, roughly a third of the cost
Fable 5 (baseline)
~3× cost
Also 61% less time and less than half the output tokens for comparable work.
Benchmarks
Sol scores vs Fable 5
Agents' Last Exam 53.6 (+13.1)
Coding Agent Index 80 (+2.8)
The Codex Video Loop
Drop in recording
screen file + natural language
→
Edit + motion
graphics & audio generated
→
Verify output
model runs real checks
Powered by ultra mode (parallel sub-agents) and Programmatic Tool Calling for long, complex tasks.
Praise
Social clips & pro-grade motion video from one recording in minutes
High marks for motion graphics, product demos, animation
End-to-end YouTube generation: script → audio → avatar → edit
Caveats
Heavy token consumption; ultra mode is costly
Even near-perfect output warrants human review
Some don't feel a big leap versus Fable
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…