Everything in the catalog

Multilingual speech recognition from audio files or the microphone.

Chat demo of the efficient Gemma 3n E4B instruction model.

Removes silent sections from an audio file.

Traces a bitmap image and outputs an SVG vector version.

Adds color to black-and-white photographs with a deep-learning colorizer.

Converts a vocal recording so it sounds like a different singer.

Makes an image inspired by an uploaded piece of music.

Runs the llama2.c small Llama model in WebAssembly with Rust Candle, generating text in the page.

Turns a portrait photo and typed text into a talking avatar video.

Labels who is speaking when in streamed audio.

Predicts training memory use for a transformer from pasted config or manual architecture values.

Edits images with Qwen Image Edit 2509 using a collection of selectable LoRAs.

Official demo for editing images by describing the change in text.

Runs several vision-language captioning models, such as DeepCaption and SkyCaptioner, on your media.

Ask questions about a chart image or have it summarized by a vision-language model.

Makes and edits images with a 6-step Qwen-Image-2.1 model and compares it to the base.

Converts an image or PDF page to Markdown with a compact OCR model.

Chat with the small Llama 3.2 3B Instruct model.

Generates protein designs from a prompt in a browser interface.

Click points on an image and get object masks from Segment Anything running in the browser.

Chat with Qwen3-VL about images or video clips.

Splits an image into separate editable layers.

Transcribes audio and attributes each passage to a speaker.

Generates video from a text prompt or an image, and can extend an existing clip.