See the tradeoffs.

Explore the comparison workflow before running your own models. Sort sample results, compare two directions and keep your notes.

Interactive synthetic results. All sample names, timings, scores and prices below are fictional.

Free sample experience

Fictional results

Scroll the table sideways to see costs, scores and your ratings. With a keyboard, focus the table and use the arrow keys.

Compare up to twoStateExample timeIllustrative costExample scoreYour rating

Side by side

Select two rows to compare the example tradeoffs. Your ratings stay on this device.

The actual local tool

Repository screenshots show the full workflow. Model names, prices and outputs are historical evidence, not current measurements or recommendations.

Historical GenMedia Benchmark results screen from its repository
Saved repository results screen, reviewed 9 November 2026. No live benchmark was run for this public demo.
View the model-picker screenshotHistorical local model picker

Explore the existing public repository. The full app runs locally with your own keys.

Run a real comparison.

Run the same prompt through two image models with fal, one at a time. Compare actual outputs, elapsed time and estimated provider cost. These results are separate from the fictional samples above.

Shared AI budget: $50/month total, $10 per project. This device: $0.50/month; this network: $1/month. No automatic retries. Inputs go to the named AI provider. Our server does not save your inputs or generated outputs. Save anything you want to keep.

Checking AI availability…

About this public version

The original app is a local tool with PGlite, ffmpeg and provider-key setup. It has a local daily budget, not a verified public multi-user spend service. This demo has synthetic table data and historical screenshot evidence. The separate public image comparison below can generate only two allowlisted models when its spending gate is enabled. Video generation, AI judging and the private local workspace are not connected.

Public tools are separate from private workspaces. AI availability and limits are shown beside generation controls. Payments, publishing and account actions are unavailable.