All GPU classes · GPU rental price
B200 rental price per hour
What an hour of B200 costs to rent right now on three sources, what the price did over the last weeks, and when renting beats buying. Read hourly from public listings; described, never forecast.
Vast.ai median of the last seven days: $8.56 per GPU-hour. Against the week before: +5 %. Against 30 days ago: +42 %.
Radar Rent Index
What the B200 rental price did
One line per source from the same hourly snapshots: the current window against the one before it, and against the first week collected. A source with too few days says so instead of showing a change.
| Source | This week | vs. last week | vs. 30 days ago | Since start | Offers | Day band $/h |
|---|---|---|---|---|---|---|
| Vast.ai medianlatest day Oct 1 | $8.56median of 7 days | +5 %flat | +42 % | +42 %since Aug 26 · 37 days on record | 15more offers than last week | $6.25–$8.13min–p75 of the latest day |
| Clore.ai mediancollecting | —collecting | —collecting | — | —collecting | —no comparison yet | — |
| RunPod fromlatest day Oct 1 | $5.98median of 7 days | 0 %flat | — | —collecting | —no comparison yet | —min–p75 of the latest day |
No day was excluded.
A day is the median of its hourly medians, counting only hours with at least five offers (verified hosts on Vast.ai, free hosts on Clore.ai). A week is the median of its days and needs at least four of them. Words follow fixed thresholds: up or down from ±8 %, sharply from ±20 %; offers up or down from ±15 %. How the index is made →
Price history
Price moves, last 30 days
Day-over-day moves of 10 % or more on a marketplace, and any change of the RunPod list price, between complete days.
- Sep 18 · Vast.ai: −12 % ($8.75 → $7.70)
- Sep 13 · Vast.ai: +13 % ($6.88 → $7.75)
- Sep 6 · Vast.ai: +25 % ($6.01 → $7.51)
Rent or buy
A single B200 has no sourced purchase price: it is a data-center part sold inside systems, so there is no honest break-even to print.
The guide: buy or rent a GPU? →What 180 GB holds
35 of 46 tracked open models fit a single B200 at 8K context, measured file by file; the largest comfortable one is Qwen3.8 Flash-Next.
Check a model in the finder →