HunyuanImage-3.0

HunyuanImage-3.0, open-sourced September 28, 2025, is Tencent's native multimodal image generation model: an 80-billion-parameter mixture-of-experts (13B active per token) that unifies multimodal understanding and generation in an autoregressive framework rather than a conventional diffusion pipeline. Tencent positions it as the largest open-source image generative model to date, with the Instruct variant adding reasoning-based prompt enhancement and image-to-image editing.

The license is the critical caveat: despite Tencent's 'open source' framing, it is a custom community license, not OSI-open — it excludes use in the EU, UK, and South Korea and requires a separate license above 100M monthly active users. At 80B parameters it also demands multi-GPU servers or managed inference, not a laptop.