Skip to content

Frequently Asked Questions

Common questions about WasiAI, organized by topic.


General

What is WasiAI?

WasiAI is a decentralized marketplace for AI models and agents built on Avalanche. It enables AI builders to monetize their work through pay-per-inference micropayments and perpetual license NFTs.

Is WasiAI live?

WasiAI is currently deployed on Avalanche Fuji testnet. Mainnet launch is planned for Q3 2025 after security audits.

Do I need crypto experience to use WasiAI?

No! You can sign up with Google, Apple, or email. A wallet is created for you automatically via Thirdweb.

What blockchain does WasiAI use?

Avalanche C-Chain. We chose Avalanche for its fast finality (~1 second), low fees, and EVM compatibility.


For Users

How do I pay for AI inference?

Click "Run Model", enter your input, and sign with your wallet. You'll pay a small fee in USDC (typically $0.001-$0.10). No gas required!

Do I need to pay gas fees?

For inference: No. The facilitator pays gas on your behalf. For license purchases: Yes, a small amount of AVAX for the transaction.

What's the difference between pay-per-inference and licenses?

Pay-Per-Inference License NFT
Pay per call One-time or monthly
No commitment Ownership (NFT)
Best for occasional use Best for heavy use
Gasless Requires gas to purchase

Can I get a refund?

Payments are final once confirmed on-chain. We recommend testing with small inputs first.

Is my data private?

Your inference inputs are sent to the model endpoint for processing. WasiAI does not permanently store your inputs. Check each model's privacy policy for specifics.

How do I get testnet funds?


For AI Builders

How do I publish a model?

  1. Connect your wallet
  2. Click "Publish Model"
  3. Follow the 5-step wizard
  4. Confirm the transaction

Your model will be live within seconds.

What pricing should I set?

Model Type Suggested Per-Inference Suggested License
Simple Classification $0.001 - $0.005 $10 - $50
Complex Analysis $0.01 - $0.05 $50 - $200
Generation $0.05 - $0.20 $100 - $500

How do royalties work?

You set a royalty percentage (0-20%) when publishing. You'll receive this percentage of every future sale, even if you transfer ownership of the model.

Can I update my model?

Yes! Publish an update with the same slug, and it becomes a new version in your model family. Reputation and revenue flow to the same splitter.

Can I change my royalty after publishing?

No. Royalties are immutable to protect all parties.

How do I receive payments?

Payments go to your model's splitter contract and are automatically distributed. You can check your pending balance and trigger distribution anytime.


Payments

What is x402?

x402 is a payment protocol that enables HTTP 402 (Payment Required) responses with cryptocurrency. It allows micropayments as low as $0.001 without gas fees for users.

Why USDC?

USDC is a stablecoin pegged to the US dollar, providing price stability. It also supports EIP-3009 (transferWithAuthorization) for gasless transfers.

What are facilitators?

Facilitators are services that execute USDC transfers on behalf of users and pay the blockchain gas fees. WasiAI supports: - Ultravioleta DAO: Free, Avalanche only - Thirdweb: $99/month, multi-chain

How is revenue split?

Payment → Splitter Contract
    ├── Marketplace (5%) → WasiAI
    ├── Creator Royalty (0-20%) → Original Creator
    └── Seller Share (remainder) → Current Owner

Are there hidden fees?

No. WasiAI charges a transparent 5% marketplace fee, visible in the smart contracts.


Licenses

What rights do licenses grant?

  • API Access (RIGHTS_API = 1): Call the inference endpoint
  • Download Access (RIGHTS_DOWNLOAD = 2): Download model artifacts

Can I sell my license?

Yes! License NFTs are standard ERC-721 tokens. You can transfer or sell them on any NFT marketplace.

What happens when my subscription expires?

You lose access to the model's capabilities. You can renew anytime to restore access.

Can licenses be revoked?

Yes, in cases of terms violation. Revoked licenses become invalid immediately.


Technical

What smart contracts does WasiAI use?

Contract Purpose
MarketplaceV3 Model registry, license sales
LicenseNFTV2 License NFT tokens
AgentRegistryV2 ERC-8004 agent identity
SplitterFactory Revenue splitter deployment
ModelSplitter Payment distribution

Where is metadata stored?

On IPFS via Pinata. Each model has a metadata URI pointing to a JSON file with name, description, artifacts, etc.

How do I integrate WasiAI into my app?

Use the inference API:

POST /api/inference/{modelId}
Content-Type: application/json
X-PAYMENT: <base64-encoded-payment>

{ "input": "your input here" }

See the API Reference for details.

Is the code open source?

The frontend and contracts are available on GitHub. See github.com/ferrosasfp/wasiai.


Security

Are the smart contracts audited?

Not yet. A security audit is planned before mainnet launch.

What security measures are in place?

  • Ownable2Step: Two-step ownership transfer
  • ReentrancyGuard: Protection against reentrancy
  • Pausable: Emergency stop functionality
  • Timelocks: 24-hour delay for critical changes
  • Nonce tracking: Replay attack prevention

What if my wallet is compromised?

For agents and models, wallet changes have a 24-hour timelock. This gives you time to detect and cancel unauthorized changes.

How do I report a security issue?

Email security@wasiai.com. Please do not disclose vulnerabilities publicly.


Troubleshooting

"Insufficient USDC Balance"

Get testnet USDC from faucet.circle.com.

"Transaction Failed"

  • Check you have AVAX for gas
  • Ensure you're on Avalanche Fuji network
  • Try reconnecting your wallet

"Signature Rejected"

  • Make sure you're on the correct network
  • Try disconnecting and reconnecting your wallet
  • Clear browser cache and try again

"Model Not Appearing"

  • Wait 1-2 minutes for indexing
  • Refresh the catalog page
  • Check the transaction on Snowtrace

"Inference Timeout"

The model may be loading (cold start). Wait a few seconds and try again.


Contact

How do I get help?

How do I report a bug?

Open an issue on GitHub with: - Steps to reproduce - Expected behavior - Actual behavior - Screenshots if applicable