01AGPU cloud · accelerator ledger· manual survey 2026-10-14 · automated rows 2026-10-14
H200 rental prices
What capacity costs. Compare the active venue stack for H200, the provider-class spread, and the long-run price direction where history is available.
H200 · reference price · Typical — median of 6 specialist clouds
$5.35/GPU·hr
Cite — copy a citation for the H200 reference price
Copy-ready citation
H200 reference price — Typical — median of 6 specialist clouds: $5.35/GPU·hr, observed between Oct 11 and Oct 14, 2026 (posted list/ask prices, not transactions). Cite as: AI Compute Tracker, ACT.H200.SPEC-MED@primary-on-demand-v1, https://aicomputetracker.com/gpu-cloud/h200/, AI Compute Tracker (aicomputetracker.com), CC BY 4.0; third-party terms apply to data from GPU cloud pricing pages, RunPod, accessed <date you visit>
Replace “<date you visit>” with your access date.
0.0%
vs Oct 13, 2026 · like-for-like
the 5 venues quoted in both; Together AI added
- Marketplace floor · n=1 · Vast.ai
- $3.68/GPU·hr
- Typical — median of 6 specialist clouds
- $5.35/GPU·hr
- Big-cloud list price · n=3
- $7.91/GPU·hr
H200 · Marketplace floor
$3.68/GPU·hr
Vast.ai · n=1
H200 · market coverage
11 venues
All-market access $3.68 → $10.60
H200 · Big-cloud list price
$7.91/GPU·hr
AWS (p5en) · n=3
H200 · specs and price per PFLOP-hour
Spec sources on /hardware →Spec-sheet row: H200 SXM, reviewed 2026-10-15 against vendor documents; est marks a figure that review did not confirm. Dense tensor rates only, one precision per column — never a sparse figure.
- Dense BF16
- 989TFLOPS
- Dense FP8
- 1,979TFLOPS
- Dense FP4
- not listed on the spec sheet
- Memory
- 141GB
- Bandwidth
- 4.8TB/s
- TDP
- 700W
- $ / PFLOP·hr · BF16
- $4.03
$4.03 per dense BF16 PFLOP-hour: Hyperstack's $3.99/GPU·hr — the lowest specialist on-demand list price, observed — ÷ 0.989 PFLOPS dense BF16.
● Market-implied term structure
H200 · market-implied forward curve
Kalshi-implied · Oct avg anchor $5.82 (quality-passed · $5.63–6.09) and monthly-average tenors through Aug ’27 — Aug ’27 ≥$1.66 (thin) — no carry shown: a thin end compares estimates
More actions for H200 · market-implied forward curve
Data — endpoints, license, and freshness behind H200 · market-implied forward curve
Full forward-curve snapshot: instrument-specific anchors, monthly tenors, methodology identity, and per-point ladder evidence.
curl -s https://aicomputetracker.com/forward-curves.jsonSpreadsheet twin of /forward-curves.json: one row per tenor with unit-encoded per-chip columns, quality, and ladder evidence.
curl -s https://aicomputetracker.com/curves.csvNo API key, CORS open (Access-Control-Allow-Origin: *) — fetch it from a browser, a notebook, or a cron job.
The site’s compilation is licensed CC BY 4.0 — attribution for this chart’s data “AI Compute Tracker (aicomputetracker.com), CC BY 4.0; third-party terms apply to data from Kalshi (KalshiEX LLC)”. Third-party source data — such as Kalshi market data, Vast.ai marketplace data, and providers’ list prices — remains subject to its sources’ terms.
Market snapshot
Cite — copy a citation for H200 · market-implied forward curve
Copy-ready citation
Cite as: AI Compute Tracker, “H200 · market-implied forward curve”, market snapshot as of 2026-10-16, https://aicomputetracker.com/gpu-cloud/h200/#forward-curve, Data: Kalshi (KalshiEX LLC), AI Compute Tracker (aicomputetracker.com), CC BY 4.0; third-party terms apply to data from Kalshi (KalshiEX LLC), accessed <date you visit>
Replace “<date you visit>” with your access date.
Embed — copy an iframe snippet for H200 · market-implied forward curve
Copy-paste iframe
The frame keeps this chart’s own freshness stamp, any third-party source credit, and the attribution line; the source line under it links back here with the data’s licence terms. It updates when the site redeploys.
Market depth
Preserved event ladders for 1 visible chip
- Quality-passed
- 4/12
- Closed brackets
- 5/12
- Volume
- 222.3K
- Open interest
- 65.8K
- Last record edit
- 2026-09-09 · 17:33Z
Tap point to pin← → tenor · ↑ ↓ chip · Home / End · Enter pin · Esc close
Data: Kalshi (KalshiEX LLC) — contract ladders referenced to the Ornn hourly index, snapshot OCT 16 2026 · 16:54 UTC. Nothing is smoothed: each monthly value is its own ladder’s estimate. The lines and bid/ask bands join only quality-passed points; gaps are hollow at their level, unplotted without one; no forward point is current spot or executable.
Exact values for H200 · market-implied forward curve
| Tenor | H200 implied $/GPU·hr | Status |
|---|---|---|
| Anchor | $5.82 | quality-passed · $5.63–6.09 |
| Oct ’26 | $5.82 | quality-passed · $5.63–6.09 |
| Nov | $5.99 | quality-passed · $5.50–6.49 |
| Dec | $5.77 | quality-passed · $5.61–5.93 |
| Jan ’27 | $1.72–6.31 | thin · $1.72–6.31 |
| Feb | ≤$5.44 | thin · ≤$5.44 |
| Mar | ≤$5.88 | thin · ≤$5.88 |
| Apr | unbounded | thin · unbounded |
| May | ≥$2.19 | thin · ≥$2.19 |
| Jun | unbounded | thin · unbounded |
| Jul | unbounded | thin · unbounded |
| Aug ’27 | ≥$1.66 | thin · ≥$1.66 |
Comparable price history
Loading evidence ledger…
H200 · daily venue-class bands
4 archived observation days — “Big-cloud list price”, “Typical — median of specialist clouds” and “Marketplace floor” per GPU-hour, each with its venue count (n)
More actions for H200 · daily venue-class bands
Data — endpoints, license, and freshness behind H200 · daily venue-class bands
The complete GPU price-history ledger: every series and dated observation, with billing, comparison, quality, and raw-evidence fields.
curl -s https://aicomputetracker.com/price-history.jsonPer-accelerator shard of the price-history ledger — the same series and point shapes filtered to one chip.
curl -s https://aicomputetracker.com/price-history/h200.jsonNo API key, CORS open (Access-Control-Allow-Origin: *) — fetch it from a browser, a notebook, or a cron job.
The site’s compilation is licensed CC BY 4.0 — attribution for this chart’s data “AI Compute Tracker (aicomputetracker.com), CC BY 4.0; third-party terms apply to data from Vast.ai, Amazon Web Services, Microsoft Azure, Google Cloud, Oracle Cloud Infrastructure, Vultr, GPU cloud pricing pages, RunPod, Voltage Park”. Third-party source data — such as Kalshi market data, Vast.ai marketplace data, and providers’ list prices — remains subject to its sources’ terms.
Live quotes
Cite — copy a citation for H200 · daily venue-class bands
Copy-ready citation
Cite as: AI Compute Tracker, “H200 · daily venue-class bands”, ACT.H200.HYP-FLOOR@primary-on-demand-v1, ACT.H200.SPEC-MED@primary-on-demand-v1, ACT.H200.MKT-FLOOR@marketplace-access-v1 (posted list/ask prices, not transactions), live quotes as of 2026-10-14, https://aicomputetracker.com/gpu-cloud/h200/#class-bands, AI Compute Tracker (aicomputetracker.com), CC BY 4.0; third-party terms apply to data from Vast.ai, Amazon Web Services, Microsoft Azure, Google Cloud, Oracle Cloud Infrastructure, Vultr, GPU cloud pricing pages, RunPod, Voltage Park, accessed <date you visit>
Replace “<date you visit>” with your access date.
Embed — copy an iframe snippet for H200 · daily venue-class bands
Copy-paste iframe
The frame keeps this chart’s own freshness stamp, any third-party source credit, and the attribution line; the source line under it links back here with the data’s licence terms. It updates when the site redeploys.
Daily class aggregates from published Neon quote snapshots, archived through 2026-10-14; their citable series IDs, with each comparison-policy version, are under Cite. A day or class without a comparable quote stays a gap — never interpolated; a median on fewer than 3 venues is drawn as its range (a vertical bar), never as a point on the line, and the tooltip and table view print that range.
H200 · current published quotes
Hourly prices retain each venue’s billing model, cheapest first. Observation dates vary by row: unavailable sources retain their last valid quote, never a new observation date. All rows: Provisional — Dated source-page observation; the raw payload is not yet pinned to the archive.
| Provider | Class | Billing | $ / GPU·hr | Notes | Evidence | Observed |
|---|---|---|---|---|---|---|
| Vast.ai | Marketplace | Marketplace / variable availability | $3.68 | Marketplace — live asks vary | Source — Vast.ai marketplace search API, Vast.ai | |
| Hyperstack | Specialist | On-demand | $3.99 | On-demand table; reserved and spot rates are separate. | Source — Hyperstack published pricing, Hyperstack | |
| Crusoe | Specialist | On-demand | $4.29 | GPU instance on-demand table; managed inference is a different product. | Source — Crusoe published pricing, Crusoe | |
| Modal | Specialist | Serverless / per-second | $4.54 | GPU-only rate $0.001261/second × 3,600; CPU, memory, and other resources billed separately. | Source — Modal published pricing, Modal | |
| RunPod | Specialist | On-demand | $5.29 | Secure Cloud selected; community, serverless, and cluster prices differ. | Source — RunPod published pricing, RunPod | |
| Nebius | Specialist | On-demand | $5.40 | On-demand rate effective October 1, 2026. Preemptible starting rates vary with capacity. | Source — Nebius published pricing, Nebius | |
| Together AI | Specialist | On-demand | $5.99 | GPU Clusters hourly on-demand rate; dedicated inference and preemptible rates differ. | Source — Together AI published pricing, Together AI | |
| CoreWeave | Specialist | On-demand | $6.31 | Whole-instance on-demand price $50.44/hour divided by 8; single-GPU inference is a separate offering. | Source — CoreWeave published pricing, CoreWeave | |
| AWS (p5en) | Hyperscaler | On-demand | $7.91 | Source — AWS EC2 pricing, AWS (p5en) | ||
| Oracle OCI | Hyperscaler | On-demand | $10.00 | Published USD per GPU-hour; multiply by eight for the full bare-metal instance. | Source — Oracle OCI published pricing, Oracle OCI | |
| Azure (ND v5) | Hyperscaler | On-demand | $10.60 | Source — Azure Retail Prices API, Azure (ND v5) |
Archived list prices for H200: 50 quarterly captures from 10 providers, 2024Q4 to 2026Q3, on the archived prices page.
H200 rental price — common questions
Generated from the current datasets — each answer carries its own stamp and updates when the data refreshes.
How much does an H200 cost to rent?
As of the Oct 14, 2026 survey: the big-cloud list price is $7.91/GPU·hr; the typical price (median of 6 specialist clouds) is $5.35/GPU·hr (includes RunPod, last observed Oct 11, 2026); the marketplace floor is $3.68/GPU·hr (Vast.ai, last observed Oct 12, 2026). All figures are posted list/ask prices, not transactions.
Where is H200 cheapest to rent?
As of the Oct 14, 2026 survey, Hyperstack lists the cheapest on-demand price at $3.99/GPU·hr (specialist class). Separately, Vast.ai lists the marketplace floor at $3.68/GPU·hr (last observed Oct 12, 2026) — variable availability, often interruptible or community hardware, so it is not comparable with on-demand list prices. These are posted list/ask prices, not transactions.
Where does the market expect H200 prices to go?
Market snapshot OCT 16 2026 · 16:54 UTC: Kalshi contract ladders price H200 at $5.82/GPU·hr for Oct ’26 but give no reliable level by Aug ’27 (bids put the median at $1.66 or more, no ask bounds it). Thin markets this session — no reliable front-to-back move: the Aug ’27 point is thin. Forward tenors are market-implied monthly averages referenced to the Ornn hourly index; they are indicative, not executable quotes, and no forward point is current spot.
Background reading: NVIDIA H200 on AI Wiki. AI Wiki is run by the same team as AI Compute Tracker.