InfoVis: 9B unified model tops open-source slide generation and editing, trained on 13M records
Jiahao Wang · hf · 2026-10-16
Researchers from CAS, BAAI, Zhejiang University and USTC released InfoVis, a unified 9B model for slide generation and editing.
- Data engine covering four slide-centric tasks: 13M+ continual-pretraining records and 1.9M SFT records
- Introduces VisSlideBench, a benchmark of 1,527 manually screened instances with multidimensional evaluation; automatic rankings align with blind human ratings
- Achieves top open-source performance in text rendering, instruction following, visual grounding, design, and editing
Code is open-sourced on GitHub (VectorSpaceLab/InfoVis).
More from Multimodal
- Alibaba open-sources CommerceGen-Studio for commercial-grade e-commerce images — alibabagroup · 2026-10-16
- EVA-Cap: Fudan and OPPO align video captioning with atomic audiovisual events via GRPO — Fudan-University · 2026-10-16
- SCOPED: skill-decoupled expert distillation beats mixed-capability RL across 8 benchmarks — Yuqi Xiong · 2026-10-16
- ComfyUI v0.41.0 ships day-one FLUX.3 Image support with sampler presets — lmoroney · 2026-10-16
- Krea2 Realism returns as a full checkpoint with natural skin textures and usage tips — rynaleopard · 2026-10-16
- Removing context conditioning makes image conditioning stronger in flow-guided video generation — pixlpa · 2026-10-16