01
Freeze the prompt
Lock your prompt into an immutable capsule. It can never be edited — so every future run stays perfectly comparable.
Immutability archive for LLMs
Freeze a prompt. Replay it across every model — and across time.
Lock prompts into immutable capsules, run them against hundreds of models through OpenRouter with your own key, and archive every output, cost and latency — a reproducible chronicle of how AI evolves.
Hundreds of models, one archive — via OpenRouter
Three steps to a permanent benchmark
No setup, no infrastructure. Just your prompt, your key, and an archive that never forgets.
02
Run the models
Fire the exact same prompt at hundreds of models through OpenRouter, using your own API key.
03
Compare across time
Re-run the capsule months later. Outputs, cost, latency and tokens sit side by side — a living record of AI progress.
A real capsule, straight from the archive
Fetched live from a public capsule — no account needed.
Built for researchers who take receipts
Every claim about model progress deserves evidence. TimeMachine keeps it.
Immutable capsules
Prompts are frozen at creation, byte for byte. No silent edits, no moving targets.
Multi-model benchmarking
Hundreds of OpenRouter models — frontier and open source — one click away.
Cost, latency & token analytics
Every run is measured and archived: spend, speed and token distribution.
BYOK privacy
Your OpenRouter key stays in your browser. We never see it, store it in plaintext, or proxy your prompts.
Fork public capsules
Remix any public capsule into your own line of research while the original stays intact.
Time-series archive
Re-run in 2027, diff against 2026. Watch models improve — or regress — on your prompts.
Your key never leaves your browser
- Your OpenRouter key is stored encrypted and never inspected.
- Requests go straight from your browser to api.openrouter.ai — no proxy, no middleman.
- The entire codebase is open source and auditable.
Start your archive today
Free, open source — browse the public archive without an account.