GPU cloud · H200 entry
Together AI: H200 price per hour
Together AI lists 1 current on-demand H200 configuration. The lowest solid one, 8× HGX at $5.99 per GPU-hour, ranks 9 of 15 providers against a median of $5.40. Read 2026-11-12.
Together AI, at a glance
- Lowest solid on-demand
- $5.99 per GPU-hour
- In that configuration
- 8× HGX
- Per server-hour
- $47.92
- Rank
- 9 of 15
- Against the median
- +11% ($5.40)
- Billing
- not stated
Entries
Every H200 row for Together AI
Per server-hour is the GPU price times the GPUs in the configuration. Chips mark marketplace figures and prices that start later.
| Configuration | Billing | Per GPU-hour | Per server-hour | Region | Read | Source |
|---|---|---|---|---|---|---|
| 8× HGX · NVIDIA HGX H200 cluster on-demand | not stated | $5.99 | $47.92 | not stated | 2026-11-12 | page |
Recorded with the prices
- Together GPU Clusters; all prices per GPU per hour (page). 8-GPU count inferred from the HGX baseboard. Column 'ON-Demand, Pay as you go'.
Position
Where Together AI sits on the ladder
The highlighted rung is Together AI. Other rungs link to their own ledger page.
Together AI is rung 9 of 15. The rung below is Nebius at $5.40, $0.59 less per GPU-hour. The rung above is OVHcloud at $6.20, $0.21 more.
Against the median of $5.40, Together AI is $0.59 over per GPU-hour, 11% above. Over a 720-hour month one GPU at Together AI's rate costs $4,312.80; at the median it would cost $3,888.00.
A rung is a provider's lowest solid price, whatever the size of the configuration, so a single-GPU price and an 8-GPU price share the scale. The table above shows the size behind each figure.
- Civo$3.49lowest
- Massed Compute$3.62+$0.13
- Hyperstack$3.99+$0.37
- Crusoe$4.29+$0.30
- DigitalOcean (Paperspace)$4.47+$0.18
- Runcrateours, live$4.67+$0.20
- RunPod$5.29+$0.62
- Median of 15 providers$5.40
- Nebius$5.40+$0.11
- Together AI$5.99+$0.59
- OVHcloud$6.20+$0.21
- CoreWeave$6.30+$0.11
- Verda$6.50+$0.20
- Amazon Web Services$7.91+$1.41
- Oracle Cloud$10.00+$2.09
- Google Cloud$10.60+$0.60
Commitments
Spot and reserved listings
Spot and reserved prices are different products. They sit apart from the on-demand statistics.
| Type | Configuration | Per GPU-hour | Per 720-hour month | Against its on-demand low | Source |
|---|---|---|---|---|---|
| spot | 8× HGX · NVIDIA HGX H200 cluster preemptible | $2.99 | $2,152.80 | -50% | page |
| reserved | 8× HGX · NVIDIA HGX H200 cluster reserved 31-90 days | $4.15 | $2,988.00 | -31% | page |
| reserved | 8× HGX · NVIDIA HGX H200 cluster reserved 3-30 days | $4.99 | $3,592.80 | -17% | page |
The comparison column divides each commitment price by this provider's lowest solid on-demand price ($5.99). Sizes can differ between rows, so read it as a guide.
Stated by the provider
What Together AI says about itself
- GPU Clusters prices are stated per GPU per hour. source
- B200 on-demand is offered on GPU Clusters. source
Only facts printed on a page we read are listed. Pricing page: https://www.together.ai/pricing
Questions
Together AI and the H200, answered
- How much does Together AI charge for an H200 per hour?
- $5.99 per GPU-hour for its lowest solid on-demand configuration (8× HGX, NVIDIA HGX H200 cluster on-demand), read 2026-11-12. It is the only current row.
- How much is an 8-GPU H200 server at Together AI?
- $5.99 per GPU times 8 = $47.92 per hour, $1,150.08 per day and $34,502.40 per 720-hour month.
- Is Together AI below the median?
- No: its lowest solid price is 11% above the median of $5.40 across 15 providers ($0.59 over).
- How does Together AI bill H200 time?
- The pages we read do not state a billing granularity for these rows.
- Does Together AI list spot or reserved H200 prices?
- Yes: spot $2.99 per GPU-hour (8× HGX, NVIDIA HGX H200 cluster preemptible), reserved $4.15 per GPU-hour (8× HGX, NVIDIA HGX H200 cluster reserved 31-90 days), and reserved $4.99 per GPU-hour (8× HGX, NVIDIA HGX H200 cluster reserved 3-30 days). These are not mixed into the on-demand statistics.
- Where do these numbers come from?
- 1 page on Together AI's own site or public price feed, each linked in the ledger above with its read date. The stored prices were last read 12 November 2026.
Disclosure
H200 Price Ledger is published by the Runcrate team, who also run Runcrate, an AI cloud. Runcrate is listed in the table for H200, marked "ours, live". It is sorted by price like every other row, with no pinned position. Other providers' prices are read from their own public pricing pages. We have no paid placement.