Data contracts / chart

GPU Rental Price per Effective FP8 PFLOP

Derived price-per-performance series for rented AI GPUs: public on-demand rental prices divided by NVIDIA peak FP8 (sparse) throughput — H100 SXM5 and H200 at 3,958 TFLOPS, B200…

Coverage window

Q4 2023 → Q2 2026 (9 rows)

Computed from the backing series at page-render time — never typed by hand.

History honesty

Observations recorded as published; no archive backfill flag on this series.

Cadence

Event-driven — curated prints land when a dated public source publishes one (per-dataset cadence weekly → annual, see recommended_cadence). The market-data agent sweeps Sundays 02:00 UTC; prints that clear its validation gate (source-excerpt verification included) apply automatically, the rest queue for human review.

Revision policy

Points change only through staged updates (market_data_staged_updates) with an audit log; the Pro export carries X-Corrections-Count.

Sources (source_name)

Silicon Analysts (derived) (9)

Known gaps

Known-gaps review pending — the coverage boundaries above are computed and current, but the curator’s qualitative gaps assessment has not been written yet. We state that rather than inventing one.

Access

Free API key

API: /api/v1/market-data/gpu-rental-price-per-tflop

Machine-readable

/api/v1/catalog/gpu-rental-price-per-tflop/contract