ArcPort

Vercel
application badge
Created by team Zmaxx on April 21, 2026
Per-API Monetization EngineUsage-Based Compute BillingGoogle

ArcPort is an API payment layer for agents on Arc. ArcPort V2 introduces session-based API payments for repeated agent usage. The core idea is simple: agents should not treat every repeated API or model call as a separate payment event. The demo shows one complete lifecycle. An operator starts with a funded Circle wallet, opens a 0.01 USDC session onchain, then an agent makes repeated paid Gemini Inference calls through signed offchain vouchers. After the calls, the session is closed once onchain. ArcPort settles cumulative spend, refunds the unused balance, and shows the full proof: open tx, close tx, calls total, cumulative spent, refund, and channel status. ArcPort has two payment modes. Charge mode is the simple path for one-off calls through Circle Nanopayments. Session mode is the differentiator: one onchain open, many signed offchain calls, one onchain close, and refund of unused budget. This matters because $0.001 API pricing breaks if every request needs its own onchain payment transaction. ArcPort focuses on the reusable payment primitive behind repeated agent API and model usage, rather than a single vertical app or a transaction-count demo. The web app is the operator control plane for wallet setup, session management, and proof. The runtime user is the agent integrating through the API layer.

Category tags:

"Circle + Arc integration is deeply embedded — not cosmetic. The session-based payment model (one onchain open, many signed offchain calls, one onchain close + refund of unused budget) is exactly the kind of innovative design that makes Arc's low-cost settlement essential. Sub-cent pricing at $0.001/call is verified, and the demo shows 56 events on Arc testnet — well above the 50-transaction threshold. The demo link points to a live hosted product at arcport.xyz, not just a video or tweet, which is a strong differentiator. The playground UI is fully functional: it shows real Gemini Inference API calls being metered and paid via signed session vouchers, with a live log, cumulative spend tracking, refund proof, and verifiable TX hashes. Two payment modes (Charge and Session) show thoughtful product design beyond a basic prototype. The "session mode" — open once onchain, call many times offchain, close once onchain — is a genuinely novel approach that solves the core problem of $0.001 pricing breaking at per-request onchain cost. Technical depth is high: offchain voucher signing, onchain session management, proof-of-spend, refund logic, and integration with Gemini inference. Business value is clear and scalable — any agent using repeated API/model calls benefits from this payment primitive. Originality is strong; this is not a payments demo but a reusable payment infrastructure layer for the agentic economy. Presentation is excellent with a polished live demo UI as the video backdrop. Overall, a standout submission."

avatar

Dharma Singh

Senior Development Manager

"Session-based API payments — open one onchain channel, run many signed offchain Gemini calls, close once with cumulative settlement and refund of unused budget. Effectively a state channel for agent inference. Clean lifecycle demo. The pattern is a known optimization but execution is solid and the lifecycle proof (open/close tx, cumulative spend, refund) is satisfying."

avatar

Naman Goyal

"The session-based payment channel is the standout idea here — open one USDC session, make repeated signed API calls with cumulative EIP-712 vouchers, close once, refund unused budget. That's a genuinely practical primitive that solves the "per-call gas kills margins" problem elegantly. 51 verifiable on-chain transactions with a custom Solidity contract, HMAC-signed webhooks, and a 260-line session handler with real error handling shows this isn't a toy. The live demo at arcport.xyz works end-to-end. Solo dev with 30 commits — impressive output. What would push it higher: organic agent traffic instead of scripted proof runs, and a video walkthrough showing the full session lifecycle in action. "

avatar

Vasu Raj Jain

Senior Software Engineer