Pricing methodology & quality gate

How we source and check compute prices, maintain strict zero-hallucination data discipline, and automate static generation without thin content.

1. Data verification discipline

Centralized hyperscalers (like AWS, Google Cloud, and Azure) publish opaque price calculators and charge outbound data egress fees ($0.09/GB) that hide the true cost of GPU workloads. Decentralized Physical Infrastructure Networks (DePIN) operate transparent market mechanics, but price formats vary across on-chain bids, job-based compute credits, and hourly reservations.

Our 4 golden rules of pricing:

•Zero invented numbers: If a price cannot be audited from an on-chain lease or public catalog, it is stored as null and prominently marked as "Not verified yet". We never extrapolate or guess.
•Mandatory audit timestamps: Every single price entry includes a visible Last checked: [date] stamp and a direct link to the primary verification source.
•Hourly normalization: Workloads charged per-token, per-frame, or through dynamic reverse-auction bidding are normalized to effective single-GPU hourly USD equivalents based on standard benchmark throughputs.
•Hyperscaler baselines: AWS EC2 prices are calculated on-demand for standard US-East-1 instances without multi-year reservation discounts to provide an honest, uncommitted comparison baseline.

2. The static quality gate

To protect user trust and uphold search engine quality guidelines, DecentralGPU implements a programmatic quality gate. We refuse to generate thin boilerplate pages or empty template shells.

Before any static route is emitted during the build process, the generator evaluates:

Network records must contain verified technical descriptions exceeding 50 characters, at least 2 distinct strengths, and 2 verified limitations.
Comparison pages must include specific, non-repetitive architectural teardowns and unique FAQ items.
Hardware specs must include verifiable memory bandwidth and compute metrics from manufacturer whitepapers.

3. Developer guide: how to add a new network or GPU

DecentralGPU is 100% statically generated from local JSON databases stored in /src/data. You do not need a database or admin dashboard to add new providers. Adding data to these JSON files automatically triggers static page generation on next build.

Step A: Add network to `/src/data/networks.json`

Append your network object. Ensure all required fields meet the quality gate criteria:

{
  "slug": "fluence",
  "name": "Fluence",
  "shortName": "Fluence",
  "headline": "Cloudless decentralized computing platform",
  "description": "Fluence is a decentralized serverless computing platform...",
  "chain": "IPC / Filecoin",
  "token": "FLT",
  "governance": "Fluence DAO",
  "workloads": ["Serverless Functions", "AI Inference", "Data Pipelines"],
  "strengths": [
    "Sub-second container initialization",
    "Deterministic execution proofs"
  ],
  "weaknesses": [
    "Primarily CPU-centric with emerging GPU support",
    "Smaller high-end accelerator pool"
  ],
  "architecture": "Aqua workflow orchestration with decentralized Kademlia network.",
  "sla": "Validator-monitored task consensus",
  "minimumCommitment": "Pay-as-you-go per execution",
  "website": "https://fluence.network",
  "affiliate_url": "https://fluence.network/?ref=decentralgpu",
  "foundedYear": 2017,
  "last_checked": "2026-10-04"
}

Step B: Add GPU to `/src/data/gpus.json`

Define hardware parameters and memory bandwidth:

{
  "slug": "l40s",
  "name": "NVIDIA L40S (48GB)",
  "shortName": "L40S",
  "architecture": "Ada Lovelace",
  "vram": "48 GB GDDR6",
  "memoryBandwidth": "864 GB/s",
  "fp16Tflops": "366 TFLOPS",
  "fp8Tflops": "733 TFLOPS",
  "interconnect": "PCIe 4.0 x16",
  "tdp": "350W",
  "typicalUseCases": ["Omniverse & 3D Workflows", "Mid-Size Model Fine-Tuning"],
  "summary": "NVIDIA L40S is a versatile universal data center GPU...",
  "buyerGuide": "Ideal for multi-modal generative AI where 48GB VRAM is sufficient..."
}

Step C: Add price entries to `/src/data/prices.json`

Link the network slug to the GPU slug. Set pricePerHourUsd: null if unverified:

{
  "networkSlug": "fluence",
  "gpuSlug": "rtx-4090",
  "pricePerHourUsd": 0.40,
  "sourceUrl": "https://fluence.network",
  "sourceNote": "Public testnet provider rate for RTX 4090",
  "billingType": "hourly",
  "isVerified": true,
  "lastChecked": "2026-10-04"
}

Step D: Build & deploy

Run npm run build. Next.js statically renders:

/networks/fluence (Network profile)
/gpus/l40s (GPU cross-network price sheet)
All associated 1-on-1 comparisons and sitemap entries

← Return to index About our mission →