Design benchmark

GPT-5.6 Sol results.

ProviderOpenAI
Official releaseJul 9, 2026 ↗
BenchmarkedJul 10, 2026
Published tests2

Model results

Solar System UI Benchmark

View full result ↗
GPT-5.6 Sol output showing a warm heliocentric solar system with labeled planets, orbit paths, and a compact control panel.
Compare with other models
Test scope and inspection guide

Motion, controls, selectable planets, visual hierarchy, and a complete browser experience.

  • Does the scene feel complete and visually coherent?
  • Do motion, navigation, and controls respond clearly?
  • Can you explore planets without instructions or repair work?
  • Does the output remain usable at the supplied viewport?

Air Traffic Control Game Benchmark

View full result ↗
GPT-5.6 Sol output showing a sector radar, aircraft tracks, weather hazard, runway, and operational side panels.
Compare with other models
Test scope and inspection guide

Aircraft routing, runway assignment, conflict logic, warnings, scoring, and escalating traffic.

  • Can a new player understand the first action quickly?
  • Do aircraft controls and runway assignments work?
  • Are conflicts, warnings, and failure states legible?
  • Does the game remain responsive as traffic develops?

Same task.
One shot. No edits.

Each model receives the same fixed instruction for a task.One recorded attempt is preserved so you can inspect the work directly.

Same taskFixed task version and output format
One attemptNo follow-up prompt or manual repair
Full resultInteractive output remains available