Introducing Idefics2: A Powerful 8B Vision-Language Model for the community

In three linesHugging Face releases Idefics2, an open-source 8B vision-language model handling images and text. Supports up to 1024×1024 pixel resolution and runs on standard hardware. Available under Apache 2.0 license with public weights and code.

Summary generated by Claude — human-verified