For indie game dev, I often need to generate images with Chinese text in them. Tried a bunch of models, and Jimeng is way more stable than most when it comes to writing Chinese characters in images. Someone reverse-engineered Jimeng’s web API and wrapped it into a free REST API, so I grabbed it and gave it a shot. It supports multiple model switching, batch generation, and a few style presets—realistic, anime, oil painting, you name it.
No API key needed, it runs off the free web channel. The catch is a rate limit of 20 images per hour, and you have to plug in a valid web session cookie—once it expires, you gotta redo it. The author also says not for commercial use, and you’ll need to check ByteDance’s terms yourself.
For someone like me who occasionally batch-generates concept art, it’s good enough, but don’t expect rock-solid stability—it’s a reverse-engineered thing, so if the API changes, you’re stuck updating. Anyone else using other domestic image gen models? Let’s chat.