US-East 42ms p50 · EU-Central 38ms p50 · AP-South 71ms p50 · ETH Block #21.4M head · SOL Block #312.1M head · Groq 750 tok/s · Together 600 tok/s · Alchemy 25ms rpc · US-East 42ms p50 · EU-Central 38ms p50 · AP-South 71ms p50 · ETH Block #21.4M head · SOL Block #312.1M head · Groq 750 tok/s · Together 600 tok/s · Alchemy 25ms rpc ·

Data Methodology

How we collect, verify, and maintain the infrastructure pricing data powering BenchNode.io.

Dataset last refreshed: June 4, 2026
51 configs across 16 providers

1. Data Collection

All pricing data is sourced directly from each provider's official public pricing pages and API documentation. We do not rely on third-party aggregators or user submissions. Each provider has a dedicated scraping pipeline that targets the canonical pricing page.

Performance metrics (latency, throughput, uptime) are sourced from provider-published SLAs and publicly available benchmark reports. Where providers do not disclose specific numbers, we use conservative industry-standard estimates and clearly mark them as indicative benchmarks.

2. Refresh Cadence

The data pipeline runs automatically every Sunday at midnight UTC via GitHub Actions. The workflow: scrape provider pricing → AI enrichment pass → commit updated JSON → trigger a fresh Cloudflare Pages build. Every page you see reflects the most recent Sunday refresh.

# .github/workflows/auto_refresh.yml
cron: '0 0 * * 0' # every Sunday 00:00 UTC

3. AI Analysis Layer

Each provider tier's verdict and ideal_use_case fields are generated by GPT-4o-mini using structured JSON output schemas. Prompts enforce citation of specific numeric values — cost-per-million-tokens, TPM/RPM ceilings, latency figures — preventing generic summaries.

AI-generated content is clearly labelled on every page. It is a synthesis of the structured data, not a substitute for reading the provider's official documentation.

4. Comparison Methodology

VS comparison pages are auto-generated using itertools combinations filtered by mutual model or chain support. A composite score ranks winners across:

  • AI Inference: cost-per-token (40%), throughput (40%), latency (20%)
  • Web3 RPC: monthly cost (30%), quota volume (50%), latency (20%)

Composite scores are indicative only. Your optimal provider depends on your specific workload shape, request volume, and latency requirements.

5. Data Sources

Provider Source Type
Groq Public pricing page
Together AI Public pricing page
Fireworks AI Public pricing page
DeepInfra Public pricing page
OpenAI Official API docs
Anthropic Official API docs
Google Official API docs
xAI Official API docs
Alchemy Public pricing page
QuickNode Public pricing page
Infura Public pricing page
Ankr Public pricing page
Chainstack Public pricing page
dRPC Public pricing page
Helius Public pricing page

6. Disclaimers

  • · BenchNode.io is not affiliated with, endorsed by, or sponsored by any listed provider.
  • · Pricing can change without notice. Always verify costs directly with the provider before committing to production workloads.
  • · Performance figures (latency, throughput) are indicative benchmarks from provider-published data, not independently measured.
  • · Some pages contain affiliate links. If you sign up through a link, we may earn a commission at no additional cost to you.
  • · AI-generated verdict and ideal_use_case text is supplementary analysis, not professional infrastructure advice.