Built for exploration

Explore 14 AI experiences from one visual hub.

Test chat, image, video, and document AI across nine model ecosystems, then benchmark compatible models side by side before you write integration code.

14 experiences. 9 model ecosystems.

Use provider-aware controls, reusable profiles, complete session archives, cost estimates, and a multi-model benchmark in one workspace.

Built in

Move from a first prompt to a repeatable comparison.

Benchmark compatible models

Send one prompt to two or three compatible playgrounds, including different models from the same family, and inspect the results side by side.

Save profiles or full sessions

Reuse settings as JSON profiles, optionally with machine-encrypted credentials, or archive non-secret settings, outputs, and generated media in a restorable ZIP session.

Measure every run

Track elapsed time, estimated Foundry cost when available, generated character counts, and returned usage or reasoning metadata.

Use provider-native features

Explore reasoning, verbosity, prompt caching, web search, WebIQ, X MCP, file and image inputs, generation controls, and OCR where supported.

Available now

Choose a focused playground or compare models across the same task.

Every playground exposes the controls and outputs that matter for its provider and modality.

Video 2 playgrounds

Create, edit, extend, or steer video with text, image, reference, and video inputs.

Document AI 1 playground

Run Mistral OCR and inspect rendered Markdown, raw Markdown, structured JSON, and page-level results.

Compare 1 experience

Load two or three supported playgrounds, reuse one prompt, and compare their outputs in parallel.

Providers on Azure

Explore nine model ecosystems without losing your workflow.

Azure OpenAI and Direct from Azure or Foundry-compatible experiences keep provider-specific endpoint patterns, model controls, and output details visible while preserving a familiar playground workflow.

Getting started

Explore, measure, and carry the useful work forward.

1

Choose an experience

Start with conversation, image generation, video creation, document understanding, or the multi-model benchmark.

2

Configure and run

Tune provider-specific settings, prompts, tools, inputs, and optional files while watching timing, cost, and metadata.

3

Compare and continue

Benchmark compatible models, then save a reusable profile or complete session without persisting secure settings.

An unhandled error has occurred. Reload 🗙