Code2Video-AA at a glance
The new benchmark, based on HeyGen's Code2Video Bench, asks 27 models to turn 114 product and brand video briefs into animated HTML; at launch, Claude Opus 5.5 led and cost per task ranged from $0.01 to $3.89.
What changed with Code2Video-AA
On October 26, 2026, Artificial Analysis launched Code2Video-AA, its independent implementation of HeyGen's Code2Video Bench. Each model receives a short design brief for a product or brand video and HeyGen's system prompt in one call, with no tools, and must return a single HyperFrames HTML composition that uses HTML for layout and GSAP for animation. The rendered videos are compared by a judge on five axes and scored as Elo, anchored to GPT-5.5 (xhigh) at 1000.
In the launch results dated October 26, 2026, Claude Opus 5.5 (max with fallback) led at 1536 Elo and was first on all five axes. Claude Sonnet 5.5 at 1443 and GPT-6 Astra (max) at 1427 were within noise of each other, and GPT-6.1 Sol (max) was fourth at 1369. The best open-weight models, DeepSeek V4.1 Flash and two MiMo-V2.6 models, scored about 1052, at least 483 points behind the leader.
Cost per task varied more than 300 times, from $0.01 for MiMo-V2.6-Flash to $3.89 for Claude Fable 5.1. At launch, GPT-6.1 Sol (max) cost $0.31 per task, about a tenth of Claude Opus 5.5 at $3.08, and Claude Haiku 5.5 (max) scored 1192 at $0.08. Of 3,078 tasks, 98.8% rendered to a video.
Where to compare Code2Video-AA
Independent leaderboards and official resources for comparing this model yourself. Rankings change often; each site shows its own current data.
Leaderboards
General leaderboards
Official resources
Why Code2Video-AA matters for businesses
For agencies producing short social or product videos, this is one of the first public comparisons of models writing animated video as code rather than generating pixels. Code-based video is easy to edit, brand, and reuse, so a mid-priced model on the cost frontier may be the practical pick; check the live leaderboard before choosing, because results change as models are added.
Sources for this story
- Code2Video-AA: how well do AI models design videos in code? (opens in a new tab)Official
Artificial Analysis · artificialanalysis.ai
- Introducing the Code2Video benchmark (opens in a new tab)Benchmark
HeyGen · heygen.com
Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.
- #artificial-analysis
- #benchmarks
- #video
- #code-generation
- #heygen
