Last Baseline

Compare a retiring model with its replacement

Everything on this page runs in your browser. How that is enforced.

1 API keys used only by this tab, sent only to the provider

· You only need the key for the provider you are testing. A restricted or temporary key works fine.

2 Models

3 Prompts

Import accepts one JSON object per line with messages (chat format, a recorded final answer is ignored), input or prompt, or a CSV with prompts in the first column.

Each prompt is sent once to each model on your key, with the provider's default sampling.

4 Results

Green and red marks show words only one model wrote. Answers vary between runs, so treat a single difference as a lead to check.

Run your whole prompt set and keep the baseline

Free

$0
  • 3 prompts per run
  • Side by side answers with word level diff
  • Automatic break checks on every answer
  • Latency and token change
Open the tool

Migration Pass

$29 once, 60 days
  • Unlimited prompts per run
  • Bulk import from JSONL or CSV, including chat logs
  • AI verdict on every pair: equivalent, improved or regressed, with the reason
  • Baseline archive: keep the retiring model's answers forever
  • Team report as one HTML file you can print to PDF
Get the Migration Pass

Always On

$12 per month
  • Everything in the Migration Pass
  • Stays unlocked for every future retirement
  • Test any new model against archives you saved earlier
  • Cancel any time
Subscribe

You pay your model provider for the API calls you run, on your own key. 14 day refund on either plan, no questions asked. Secure checkout by Stripe.