MeshInfer + MeshNativeExchange: Distributed AI Inference with Vital Billing

MeshInfer is the distributed inference layer of the Distributed Intelligence Stack. Every inference request executed on MeshInfer is metered in Vital DIU and settled through MeshNativeExchange — giving enterprises a single, auditable billing record for all distributed AI compute.

MeshInfer (available at meshinfer.ai) is the distributed AI inference platform of the mesh. It executes large language model, embedding, and vision inference workloads across the mesh network — routing requests to the best available node based on privacy tier, latency requirements, and model availability. Every MeshInfer request generates a signed Vital spend record, metered at deterministic rates per 1,000 tokens processed, and forwarded to MeshNativeExchange for enterprise settlement.

Distributed Inference Execution

MeshInfer routes inference requests across available mesh nodes, selecting the optimal node based on model availability, latency, and privacy tier.

Sub-100ms Inference Routing

The MeshInfer routing layer adds less than 100ms overhead to inference requests — maintaining real-time responsiveness for production AI applications.

Privacy-Native Inference

Confidential and Sovereign inference workloads are cryptographically isolated at the enclave level — ensuring model weights and data never co-mingle across tenants.

Vital-Metered Billing

Every MeshInfer request is metered per 1,000 tokens processed and billed in Vital DIU with privacy tier multipliers applied automatically.

MeshInfer Model Catalog & Vital Rates

Model	Use Case	Base Rate	Privacy Tiers
mesh-llm-7b	General-purpose, low-latency	2.0 DIU / 1K tokens	Standard, Confidential
mesh-llm-70b	Enterprise reasoning & analysis	2.0 DIU / 1K tokens	All tiers
mesh-llm-vision	Multimodal: text + image	3.0 DIU / 1K tokens	Standard, Confidential
mesh-embed-large	Semantic embeddings & RAG	0.5 DIU / 1K tokens	All tiers
mesh-code-expert	Code generation & review	2.5 DIU / 1K tokens	Standard, Confidential

Privacy multipliers (1.8× Confidential, 3.2× Sovereign) applied on top of base rates. Rates are deterministic and signed at the protocol level.

Integration Examples

Single inference request (Sovereign tier)

meshinfer-complete.ts

import MeshInfer from "@meshinfer/sdk";

const mi = new MeshInfer({ apiKey: process.env.MESHINFER_KEY });

// Run a sovereign-tier inference request
const response = await mi.complete({
  model: "mesh-llm-70b",
  messages: [
    { role: "system", content: "You are an enterprise settlement analyst." },
    { role: "user", content: "Summarize Q2 Vital redemption activity." },
  ],
  privacyTier: "sovereign",  // 3.2× Vital multiplier — data stays in jurisdiction
  maxTokens: 2048,
});

console.log(response.content);
console.log(response.usage.vitalConsumed);   // DIU debited
console.log(response.usage.tokensGenerated); // tokens produced
console.log(response.settlementRef);         // immutable billing record

Batch embedding with Vital budget cap

meshinfer-batch.ts

// Batch inference with budget cap
const batch = await mi.batch({
  model: "mesh-embed-large",
  privacyTier: "confidential",
  inputs: documents,           // array of strings
  vitalBudget: 50000,          // hard cap; returns partial results if exceeded
});

console.log(batch.embeddings);
console.log(batch.vitalConsumed);
console.log(batch.processedCount);

MeshInfer → MeshNativeExchange Billing Integration

MeshInfer billing is fully automatic. At each billing cycle boundary, MeshInfer forwards a signed, aggregated spend record to MeshNativeExchange covering all inference activity: model used, tokens processed, privacy tier applied, Vital consumed per request, and the cumulative total for the period. Enterprises view this in the Enterprise Dashboard and redeem via the Vital Redemption flow. No manual reconciliation required.

DOSFI Integration 20-Node Test Mesh Vital Spend Compute Settlement Enterprise Onboarding Vital Logs