LuxTTS is a compact but powerful voice cloning model that turns text into natural sounding speech at very high speed, making it ideal for product teams that need scalable, real time synthetic voices without heavy infrastructure or complex licensing.

Sam Audio Large is an AI model that lets you isolate any sound from complex audio with text, visual or time-based prompts, transforming audio cleaning, music production and content workflows across media and enterprise use cases.

OneReward is a new reinforcement learning from human feedback framework for image models that uses one powerful vision language reward model to guide many different image editing tasks, delivering more consistent quality and business value than task specific fine tuning approaches.

Qwen Image 2512 is a next generation text to image model that delivers highly realistic people, detailed natural scenes, and crisp text, making it a strong option for marketing, product, and design teams that need reliable, scalable image generation for real world use cases.​

Imagine you made a fun video in English and now want kids in Spain, Brazil, and Japan to enjoy it as if you spoke their language from the start. ElevenLabs Dubbing is like a smart magic translator that listens to your voice, understands what you say, translates it, and then…

Wan Move is an emerging motion controllable video generation framework that lets teams draw precise motion paths for objects and cameras, then automatically produce short, high quality videos that follow those paths with minimal model changes and open licensing, making it a powerful building block for creative, commercial, and product…

Imagine you recorded a friend talking in a noisy kitchen with an old phone. The voice sounds small and cloudy, and you can hear the room more than the person. Nova SR is like a magic cleaner that takes this messy sound and makes the voice big, clear and easy…

ElevenLabs voice changer turns any spoken audio into a new, natural sounding voice while keeping emotion, timing, and delivery intact, making it a powerful tool for creators, brands, and developers across content, gaming, learning, and customer experience workflows.​

GLM Image is a new generation text-to-image model that combines an auto-regressive brain with a diffusion decoder to create sharper, more controllable visuals from natural language prompts and reference images. It is designed for information-dense scenes, precise text in images, and brand-level visual consistency, which makes it especially attractive for…

Deepfilternet 3 is a compact deep learning model that delivers strong real time noise suppression for speech, making calls, streams and recordings clearer without expensive hardware or heavy compute overhead.​