The playground

Streaming chat with a model picker, system prompt, sampling controls, a thinking toggle, image/audio/video attachments for multimodal models, and inline speech playback.

Turn on the logprobs viewer to see each token's probability — useful for debugging refusals, confidence, and sampling.

Conversations persist with AI-generated titles, so you can pick up where you left off. Browse them under Dashboard → Chats.

A server-side tool-calling chatbot. It can search the web, scrape pages, browse, and generate images, video, music, and voice — picking tools and skills on its own.

Hands-free: your microphone streams to speech-to-text, into the streaming agent, and back out as speech — a spoken conversation with tools.

Record or upload audio and get a transcript, with timestamps when the model supports them.

Synthesize speech from text with a selectable voice.

Manage a private library of voice samples, then generate multi-speaker dialogue in those voices.

Turn a description and optional lyrics into a finished track — with cover art.

Denoise and clean up a noisy recording.

Text-to-image and image editing.

Generate clips from a prompt, or animate a still with image-to-video.

Turn a single image into a downloadable 3D mesh.

Fetch any URL as clean Markdown.

Watch async jobs and workflow runs execute live, including an SVG-rendered DAG of each run.

A visual builder for multi-step pipelines — drag nodes, wire outputs to inputs, run, fork, and rerun a single step.

Assemble multi-segment narrated episodes from text — generate, retrim, reorder, and regenerate each segment.