Together AI

AI cloud for GPU clusters and model inference

Mid-size cloudEuropeInference

Together AI sells self-serve GPU clusters with managed Kubernetes or Slurm on InfiniBand, alongside serverless and dedicated model inference. It publishes on-demand (H100 $3.99, B200 $8.19 per GPU-hour), preemptible and short-term reserved prices for 3–30, 31–90 and 91–180 days, for example H100 at $3.19 on a 91–180 day term. Dedicated inference is listed at $5.49 per H100 GPU-hour and $8.99 per B200 GPU-hour.

Best for: Short-term reserved clusters and teams that train and serve models on one platform.

ClusterMAX 3.0: Bronze

✓ Publishes prices
✓ Serverless inference
– AMD GPUs
✓ European data centers

Visit Together AI

Facts checked November 9, 2026 against the sources listed below. Together AI does not pay to appear here.

1 Legal nameTogether AI

2 HeadquartersNot disclosed

3 Founded2022

4 Funding or listingPrivate. $800M Series C at $8.3B post-money, led by Aramco Ventures (Jul 2026).

5 How you rentOn-demand, Reserved clusters, Serverless inference

6 Minimum commitmentNone for on-demand clusters (hourly, terminate anytime); reserved terms from 3 days to 6 months, paid upfront

7 InterconnectInfiniBand (no speed stated)

8 OrchestrationKubernetes (managed), Slurm (on Kubernetes)

9 GPUs offeredGB300 NVL72, GB200 NVL72, B300, B200, H200, H100 SXM

10 RegionsUS; Europe (France, Netherlands, Sweden, Romania)

11 CertificationsNot disclosed

12 Customers and partnersCursor, Cognition and Decagon reported as customers (Series C coverage); Pika and Krea quoted on its GPU Clusters page.

Published prices

On-demand list prices per GPU-hour, as published on November 9, 2026. Reserved and negotiated prices are usually lower.
GPU	Price	Basis	Source
H100 SXM	$3.99	on-demand GPU cluster, pay as you go (HGX H100)	Pricing page
H200	$5.99	on-demand GPU cluster, pay as you go	Pricing page
B200	$8.19	on-demand GPU cluster, pay as you go	Pricing page
B300	$9.99	on-demand GPU cluster, pay as you go	Pricing page

Where we recommend them

Best for B200 and GB200 access, an alternative.
Best for inference, our pick.

Worth knowing before you sign

Downgraded from Silver to Bronze in ClusterMAX 3.0 for reliability: health checks drained nodes mid-run, and its Prince George site (owned by IREN) was described as unreliable.
GB200 and GB300 NVL72 have no public on-demand rate; reserved terms over 180 days are contact sales.
No security certifications were readable on its trust center.

Go to Together AI's site

Sources

Work at Together AI? Send corrections with a source to editors@bestneoclouds.com. We fix verified errors within five business days.

Compare with

CoreWeave GPU cloud built around large InfiniBand training clusters (NASDAQ: CRWV)
Nebius Full-stack AI cloud with published on-demand GPU prices (NASDAQ: NBIS)
Crusoe AI cloud with self-serve NVIDIA and AMD GPUs on its own data center campuses
Nscale European AI cloud selling GB300 and B200 capacity (S-1 filed for NYSE: NSCL)
Runpod Self-serve GPU cloud with per-second Pods, Instant Clusters and serverless endpoints