About
inferenstack
The market read on AI inference.
inferenstack is an inference-market terminal: a daily, honestly-labeled read on what it costs to run AI — token prices across providers, GPU rental across clouds, and the buy-vs-rent-vs- API-vs-build math — with free interactive charts as the front door and time-depth behind a gate.
The differentiator isn't coverage, it's honesty: every number carries its provenance, comparisons never mix incompatible things (tier, price basis, unit), and when we don't know something — measured latency, uptime — we say so instead of inventing a score. The forward-only daily archive is the part nobody can backfill.
Primary sources only. No fabricated numbers, ever.
Methodology
How every figure is computed, measured, or archived — and the caveats we always show.
Provenance & labels
What measured / estimated / vendor-claimed / community-reported mean, and what we refuse to claim.
Roadmap
What's shipped, what's next, and what's honestly missing until we have the data.