The Proof Batch
Gas cost vs. number of AI inference proofs under three aggregation strategies: naive per-proof (linear), SnarkPack O(log N), and ZK recursion (flat). Drag the slider to find the crossover where each strategy wins.
Archive
138 artifacts, grouped by date.
Gas cost vs. number of AI inference proofs under three aggregation strategies: naive per-proof (linear), SnarkPack O(log N), and ZK recursion (flat). Drag the slider to find the crossover where each strategy wins.
A data market can't pay you for your data's value AND protect it with one privacy budget ε. Drag ε: the worst-case leakage bound hits ≈100% by ε=8 — exactly where from-scratch models finally become usable. Real attacks sit far below the bound, and that gap is your only cover.
Every route to trustworthy AI inference, priced on one log axis: TEE attestation at 1.07×, optimistic re-execution at 2–4×, zkML and FHE at 10³–10⁵×, verifiable FHE beyond. Hover, tap, or arrow through the bars to see what each overhead actually buys — and whom you still trust.
Four guardrail tiers for on-chain AI agents: probabilistic classifiers through Lean 4 theorem provers. Select an attack scenario to see which tier catches it — and which let it through.
An on-chain agent is told to pay an invoice, but one MCP tool hides an instruction in its description that redirects the transfer to an attacker. Pick a defense layer — model, tool, client, policy, custody — and run it; alignment alone refused under 3% of these in MCPTox.
Pipeline-parallel decentralised post-training, split across four nodes. Click any internal stage to place the compromised node — the task-vector injection indicator and attack metrics update to show 94% ASR collapsing alignment from 80% to 6%.
Real reasoning-token embeddings cluster tightly in projected 3D space; phantom billing tokens scatter as noise outside the CoIn detection boundary. Toggle inflate to inject phantom tokens and watch 94.7% get flagged outside the wireframe sphere. Drag to orbit.
Three agentic payment protocol architectures—AP2, TessPay, and On-Chain ERC-4337—shown as flow stages. Toggle Attack Mode to highlight where prompt injection compromises each design. Click a column to inspect protocol details.
Compare Chainlink's push oracle price against Pyth's pull price and live spot. Drag the slider to see exactly when the 0.5% deviation threshold triggers an on-chain update.
Where reliability comes from in on-chain LLM trading agents, over DX Terminal Pro's production run: model upgrades move swap success 87% to 96%, but the operating layer takes the same weights to 99.9%. Switch panels for the prompt fixes that erased three failure modes. Tap or arrow the bars.
The six stages of an ERC-4337 UserOperation — from construction by an AI agent to paymaster settlement — with the actor, failure modes, and gas cost at each step.
Under secure two-party inference, a transformer's cost is eaten by Softmax and GELU, not the matrix multiplies. Stacked per-operation communication and interaction rounds for one BERT-base block across Iron, BumbleBee, and Nimbus — the nonlinear activations are 80–95% of the bill.