YOUR KEYS · YOUR REQUEST · YOUR VERDICT

Don’t believe the headline. Run the evidence.

Use your own OpenAI or Anthropic account, model and request. Compare direct cold, direct warm, VARION cold and VARION warm calls, then inspect prompt hashes, provider/model identity, fresh response IDs and provider-reported economics.

The world’s first publicly verifiable zero-loss AI cost optimizer.Up to 64.7% verified warm-cache cost reduction in a four-case Anthropic staging benchmark.

Challenge VARION privately

Keys are used only for this live test and are never written to the report or database.

4 provider calls
Privacy: VARION stores only hashes, usage, costs and verification status. Full prompt content, generated answers and API keys are excluded from reports.

Verification result

Results appear here after the four calls finish.

NOT RUN
🛡️Zero-Risk Proof Lab

Use a long repeated prefix for the clearest warm-cache measurement.

Do not trust the browser. Run the verifier locally.

Run the comparison from your own computer instead of trusting the browser dashboard. The verification source is public; VARION’s proprietary optimisation engine remains private.

COMPLETE TOOLKIT

Windows, Python, Node, Jupyter and cURL

Includes source, instructions and SHA-256 checksums.

Download verifier toolkit
WINDOWS

One-click verifier

Runs the open-source Python verifier and writes a local JSON report.

Download Windows BAT
SOURCE

Inspect every verification step

Download Python, Node.js, Jupyter Notebook, cURL and checksum files separately.

PythonNode.jsJupyterChecksums

What the proof verifies

Prompt-content integrity

The exact locally sent JSON body is SHA-256 hashed and compared with the hash reported by VARION before provider cache metadata is added.

Same provider and model

The selected provider and model are used for both direct and VARION calls. Zero-Loss Cache never routes to another model or provider.

Fresh generation

Distinct provider response IDs and real provider usage indicate that each response was newly generated rather than substituted from a semantic answer cache.

Provider-reported economics

The report separates direct cold, direct warm, VARION cold and VARION warm usage and costs. It never adds context reduction to cache savings.

Definition: “Zero-risk” applies only to Zero-Loss Cache Mode. VARION may add provider-supported cache metadata but does not rewrite/delete prompt content, switch provider/model or substitute a stored response. The world-first statement is based on VARION’s review of publicly documented competing products as of 4 August 2026. Results vary.