Early Access Open

Know which AI wins before you waste the prompt.

Paste your task once. Get a scored head-to-head between GPT and Claude in 30 seconds. Save the winner as a routing card you keep forever.

  • Scored battles
  • Save results as personal routing
  • Your own API keys

Five tabs. One question. No answer.

Power users paying $40+ a month for AI still copy-paste the same prompt into three windows and guess. That guessing is the problem AIToolRoulette solves.

You already pay for two AI subscriptions.

Most users stacking ChatGPT Plus and Claude Pro have no system for deciding which one to open first.

Manual comparison eats the time AI was

Copy-pasting one prompt across three tabs, then judging outputs yourself, is a ten-minute ritual dressed up as a shortcut.

One battle. One winner. One saved card.

Run a scored head-to-head in 30 seconds. Save the result as a routing card. Next time you face the same task type, the answer is already there.

Product

Everything you need to stop second-guessing your AI stack.

Every feature maps to a real step in the battle-to-routin workflow, nothing decorative.

Parallel Head-to-Head Battles

Your prompt hits GPT-5.4 mini and Claude Haiku 4.5 simultaneously, so results land at the same time with zero sequential bias.

Scored Battle Cards

Each output gets a 1-5 score on accuracy, tone, and brevity from a judge LLM, then a combined winner badge you can read at a

Side-by-Side Result View

Raw model responses sit next to each other with per-dimension scores, so you compare substance, not just vibes.

Personal Routing Cards

Save any battle result as a routing card capturing task type, best model, and your notes for instant recall next time.

Model Affinity Profile

Your routing cards accumulate into a personal profile showing which model consistently wins for each task type you run.

Encrypted BYOK Key Storage

Your OpenAI and Anthropic API keys are stored encrypted server-side and never returned to the client in raw form.

How it works

From prompt to routing card in three steps.

The whole flow runs inside your browser against your own API keys, with no shared quota and no black-box scoring.

  • Paste your API keys once in Settings, then forget them
  • Type your prompt, pick your models, click Run Battle
  • Review scored results, save as a routing card
  1. 1

    Add Your API Keys

    Go to Settings and paste your OpenAI and Anthropic keys. They are encrypted immediately. A status chip confirms each connection before you run anything.

  2. 2

    Run the Battle

    Type your prompt, label the task type, and click Run Battle. Parallel calls fire at once and a judge LLM scores every output on accuracy, tone, and brevity.

  3. 3

    Save Your Routing Card

    Click Save as Routing Card on the result page. The task type and winning model pre-fill. Add a note and your affinity profile grows with every battle.

Pricing

Free while we're in early access.

Join the waitlist now and get full access to battles, routing cards, and your model affinity profile at no cost during launch.

Early accessLimited onboarding

Free Early Access

Full battle and routing card access during the waitlist period, no payment required at any step.

  • Unlimited head-to-head battles
  • Unlimited routing cards saved
  • Encrypted BYOK key storage included
  • Early-access feedback shapes the roadmap

Honest answers before you sign up.

Straight answers for operators shipping a credible first impression.

Stop guessing. Start routing.

Every prompt you run blind is a coin flip. Join the waitlist and get a scored answer instead.