# zkML cost calculator — Proof Plan

Canonical: https://proof.agiscorecard.com/guide.html
Reviewed: 2026-09-19
Author: AGI Scorecard team
Version: 1.1.0

Proof Plan estimates proving compute and onchain verification costs from your own measurements. It checks a per-job latency target and preserves model, input and verification-key hashes. It plans an integration; it does not generate or verify a zero-knowledge proof.

## Method

Estimate proving compute and one onchain gas verification per job. Compare cost and per-job latency with supplied limits, preserving hashes and verifier version.

## Workflow

1. Measure the intended model using an established prover such as EZKL.
2. Pin model, input and verification-key hashes; record tool version and statement.
3. Review the budget, then verify the actual proof with the established verifier and trusted key.

## Common mistake

Gas-only estimates can omit L2 data fees, failed proofs, setup and queues. The displayed total covers the entered model; compare it with actual measured bills.

## Worked scenarios

### Plan 100 proofs

Inspect compute, gas, latency and artifact versions together.
Fictional inputs: https://proof.agiscorecard.com/examples/baseline.json
Run: https://proof.agiscorecard.com/?scenario=baseline#workbench
- Estimated total USD: 50.666667
- Per-job latency (s): 12.025
- Budget / latency: Within / Within

### Gas price increases

A different gas assumption can break an otherwise feasible budget.
Fictional inputs: https://proof.agiscorecard.com/examples/gas.json
Run: https://proof.agiscorecard.com/?scenario=gas#workbench
- Estimated total USD: 150.666667
- Per-job latency (s): 12.025
- Budget / latency: Over / Within

### Proving takes longer

Longer proving affects compute cost and the per-job latency target.
Fictional inputs: https://proof.agiscorecard.com/examples/latency.json
Run: https://proof.agiscorecard.com/?scenario=latency#workbench
- Estimated total USD: 51.666667
- Per-job latency (s): 30.025
- Budget / latency: Within / Over

## Questions

### Does a matching file hash verify a zkML proof?

No. A hash helps identify bytes. Proof verification requires an appropriate verifier, verification key and statement. Use established proving tools and check their trust assumptions.

### Is Proof Plan free, and do I need a wallet?

All current functions are free beta. No account, wallet connection, subscription or model API key is needed. No ZK proof generation or cryptographic verification. File hashing verifies bytes only; it is not proof of model correctness.

### Can I use my own records and keep them private?

Yes. Enter records, import JSON or paste CSV into record groups. Inputs and comparison snapshots stay in this browser tab. Share-example links contain only a public scenario name. Review downloaded reports before sharing your records.

## Alternatives and official sources

- [EZKL](https://docs.ezkl.xyz/): Converts ONNX model computations into circuits and provides proof generation and verification.
- [snarkjs](https://github.com/iden3/snarkjs): JavaScript/WASM tools for established SNARK proving and verification workflows.

Use a real prover/verifier for cryptographic assurance. This site addresses planning and artifact review only; it cannot replace EZKL, snarkjs or a security review.

## Limits

No ZK proof generation or cryptographic verification. File hashing verifies bytes only; it is not proof of model correctness.
No zero-knowledge proof is generated or cryptographically verified here. A hash records bytes, not model correctness, ownership or trusted setup. Use EZKL/snarkjs and a trusted verification key for proof verification.

Cost assumes one onchain verification per job. L2 data fees, calldata, storage, setup, failed proofs and verifier infrastructure are excluded unless included in your measurements. USD/native exchange rate is supplied, not live.

Latency is per job with no queue; throughput, concurrent capacity and SLA are not certified. Artifact hashes are declared; the optional local file fingerprint can help you check a file yourself.

## Input and output

Structural schema: https://proof.agiscorecard.com/input.schema.json
The engine additionally validates ranges, dates, amount formats, uniqueness and cross-field rules.
Input JSON and CSV paste are processed locally. No model, account or wallet is needed.
Reports can be downloaded as JSON or Markdown and printed locally. Public scenario links never include user records.

## Price and license

Current functions are free beta. No paid plan or update SLA is offered.
Tool bundle license: https://proof.agiscorecard.com/LICENSE.txt
Privacy: https://proof.agiscorecard.com/privacy.html