FULL STORY
Animation Bench Launches to Test Web Animation Skills
Physera released Animation Bench v1.0, the first benchmark for web animation reproduction, with GPT-6 Astra topping the initial run before GPT-6.1 Sol took the lead a day later.
2026-10-01 ~ 2026-10-02 · 2 episodes · 6 posts
Episode 1 · Animation Bench Launches to Test Web Animation Replication (2026-10-01, 4 posts)
Physera released Animation Bench v1.0, a benchmark testing four frontier models on 48 real-world web animation tasks across 192 rebuilds. GPT-6 Astra topped the leaderboard, though motion consistency remained weak across all models.
- Animation Bench tests 4 frontier models on rebuilding real web animations — himanshustwts · 2026-10-01
- Animation Bench full leaderboard: GPT-6 Astra leads as motion scores lag — himanshustwts · 2026-10-01
- Animation Bench Debuts to Probe Frontier Models' Multimodal Animation Gaps — himanshustwts · 2026-10-01
- Animation Bench tests coding agents on rebuilding web animations, and motion is their weak spot — himanshustwts · 2026-10-01
Episode 2 · GPT-6.1 Sol Tops Animation Bench for Web Animation Reconstruction (2026-10-02, 2 posts)
Physera's Animation Bench v1.0 evaluates coding agents on 48 web animation tasks from 32 real commercial sites. GPT-6.1 Sol ranked first overall, costing only $0.46 per task, and became the first model to exceed 0.5 in motion consistency.
- GPT-6.1 Sol tops Animation Bench, first model to cross 0.5 on motion consistency — himanshustwts · 2026-10-02
- Animation Bench: GPT-6.1 Sol tops coding agents on web animation reconstruction at $0.46/task — himanshustwts · 2026-10-02