SenseNova has released the SenseNova-U1-8b-MoT series, a set of Mixture of Transformers models designed for generating and editing dense infographics. The latest release, Infographic V2, serves as a 50-step base model that is licensed under Apache 2.
- Infographic V2: 50-step base model optimized for infographic generation.
- Infographic V1: Available as an 8-step LoRA for faster generation or as a 50-step base model.
- Interleaved images model: Generates series of related images with consistent characters, fonts, and colors.
- Hardware requirements: Approximately 36 GB VRAM for full bf16, with quantized versions available for lower memory usage.
The models are available on Hugging Face and can be served via FastAPI to provide OpenAI-compatible image generation and editing endpoints.