HyperFrames, Google DeepMind & Kaggle launch Code2Video Bench for agentic video generation
Aiden_Tech_Ai · x · 2026-09-22
HyperFrames, in collaboration with Google DeepMind and Kaggle, launched Code2Video Bench, a new benchmark specifically for evaluating code-to-video generation by agents.
The thesis: the agentic video stack is being built on code-gen, but strong SWE-bench-style coding skills don't translate into videos that feel alive. The benchmark aims to give frontier labs a measurable target for agentic video tasks.
The announcement doubles as promotion for HyperFrames' own agentic video product, arguing video agents need their own eval track.
Related event: HyperFrames Partners with DeepMind and Kaggle to Launch Code2Video Bench(6 posts)→
More from Multimodal
- P2Echo: one text-prompted model segments multi-view echocardiography with high accuracy and low cost — UBC-V · 2026-11-12
- Redditor makes fast-motion short film with H3 and ships Longshot Studio v0.18.5 with face swap — R34vspec · 2026-11-12
- Reimagining Knight Rider's opening as a Netflix drama with Claude Fable 5.2 — justin_hart · 2026-11-12
- Gemini 4 Carbon builds a full 3D aquarium in Blender and Three.js in ~6 hours — sunjiao123sun_ · 2026-11-12
- Awaken Open-Sourced: Turn a Half-Duplex Speech LLM Full-Duplex With Just a Small LoRA — alexcovo_eth · 2026-11-12
- MiniMax H3 Image targets e-commerce: sketch to 4K campaign shots in ~10 seconds — MiniMax_AI · 2026-11-12