Animation Bench tests coding agents on rebuilding web animations, and motion is their weak spot
himanshustwts · x · 2026-10-01
Physera released Animation Bench v1.0, a benchmark arguing that screenshot-parity evaluations can't distinguish visually plausible output from shippable frontend reconstruction. It asks multimodal coding agents to rebuild 48 real web animations from 32 live commercial sites (192 reconstructions), scoring visual, motion, and layout fidelity.
Leaderboard (0-1 reproduction score):
- GPT-6 Astra: 0.594, 27 task wins, $3.04/task
- Claude Fable 5.1: 0.548, 7 wins, $3.89/task
- GPT-6 Sol: 0.516, 8 wins, only $0.45/task
- Claude Opus 5.5: 0.507, 6 wins, $1.12/task
The standout finding: motion scores sit below 0.5 for every model (best: 0.473), while visual scores reach 0.63-0.71 — frontier agents can copy a page's look but still struggle to reproduce scroll, drag, and WebGL-driven behavior. Full task set and side-by-side reconstructions are in the appendix.
Related event: Animation Bench Launches to Test Web Animation Replication(4 posts)→
More from coding & agent
- shadcn boosts @uisdk: one-shot UI rebuild that's instantly E2E-testable by AI — shadcn · 2026-11-16
- AI agent deletes 43,599 lines of OpenClaw test code in one PR — vincent_koc · 2026-11-16
- JPMorgan shows Jarvis, its internal agent assembly line, at Interrupt NYC — LangChain · 2026-11-16
- Dev clarifies: Apple doesn't ban feature flags, only fooling App Review — talkaboutdesign · 2026-11-16
- Antigravity Matches Claude Cowork in Tests, While Gemini Tasks' Purpose Remains Unclear — sebajun2 · 2026-11-16
- One 16GB GPU runs songs, voice, images, 720p video and local RAG — all offline and open source — llama-of-death · 2026-11-16