CIComputeRadar
Dedicated AI inference

Together AI server rental offers

Single-tenant dedicated inference GPU-hour pricing; it is deliberately separated from Together serverless token pricing.

StatusPublished data
Offers now5
API accessPublished pricing / catalog
Updated2026-09-07 21:50 UTC
Open Together AI ↗
Vultr4181Hivelocity2695UpCloud2610OVHcloud2081Gcore1816Scaleway864Sesterce719Alibaba Cloud549Cherry Servers381Servers.com291Exoscale234Spheron200Vast.ai150Ori142Verda128Timeweb Cloud100Hyperstack79Northflank57Akash Network43RunPod42bunny.net Storage30Prime Intellect21IDrive e219Replicate19Baseten18SaladCloud17Wasabi16Paperspace15Render15Thunder Compute13Modal11Hostwinds10Melbicom10Fly.io8fal.ai7Worldstream7Impossible Cloud6Limestone Networks6Serverspace6GMI Cloud5Together AI5InterServer4Railway4TensorDock4Hyperbolic3Filebase1Storj1
⌕
5 matching offers17,643 indexed · 47/47 sources with data · catalog 12:50 AM · showing 1–5
Together AIGPUGPU cluster on-demand

Together AI GPU Cluster · HGX H100

Location not specified · Published / orderability not live-verified

$3.99/ hour≈ $2,912.70/mo
CPUConfigurableDedicated cluster-node CPU depends on the selected cluster configuration; Together publishes pricing per GPU, not per fixed VM shape
RAMConfigurableSystem RAM depends on the selected cluster configuration; not published per GPU pricing row
GPU1× HGX H10080 GB VRAM / GPU
StorageConfigurableOptional managed shared storage (WEKA/VAST) can be attached to GPU clusters; capacity is configured separately
NetworkDedicated bare-metal cluster fabric · NVLink/NVSwitch inside node and high-speed InfiniBand for multi-node scale; exact per-offer port rate is configuration-specificprovider-reported network terms; port speed not reported
Available · updated 10:43 PM
Full specs
Together AIGPUDedicated inference promo

Together AI Dedicated Inference · HGX H100

Location not specified · Published / orderability not live-verified

$3.99/ hour≈ $2,912.70/mo
CPUConfigurableDedicated cluster-node CPU depends on the selected cluster configuration; Together publishes pricing per GPU, not per fixed VM shape
RAMConfigurableSystem RAM depends on the selected cluster configuration; not published per GPU pricing row
GPU1× HGX H10080 GB VRAM / GPU
StorageConfigurableOptional managed shared storage (WEKA/VAST) can be attached to GPU clusters; capacity is configured separately
NetworkDedicated bare-metal cluster fabric · NVLink/NVSwitch inside node and high-speed InfiniBand for multi-node scale; exact per-offer port rate is configuration-specificprovider-reported network terms; port speed not reported
Available · updated 10:43 PM
Pricing noteFull specs
Together AIGPUGPU cluster on-demand

Together AI GPU Cluster · HGX H200

Location not specified · Published / orderability not live-verified

$5.99/ hour≈ $4,372.70/mo
CPUConfigurableDedicated cluster-node CPU depends on the selected cluster configuration; Together publishes pricing per GPU, not per fixed VM shape
RAMConfigurableSystem RAM depends on the selected cluster configuration; not published per GPU pricing row
GPU1× HGX H200140 GB VRAM / GPU
StorageConfigurableOptional managed shared storage (WEKA/VAST) can be attached to GPU clusters; capacity is configured separately
NetworkDedicated bare-metal cluster fabric · NVLink/NVSwitch inside node and high-speed InfiniBand for multi-node scale; exact per-offer port rate is configuration-specificprovider-reported network terms; port speed not reported
Available · updated 10:43 PM
Full specs
Together AIGPUGPU cluster on-demand

Together AI GPU Cluster · HGX B200

Location not specified · Published / orderability not live-verified

$8.19/ hour≈ $5,978.70/mo
CPUConfigurableDedicated cluster-node CPU depends on the selected cluster configuration; Together publishes pricing per GPU, not per fixed VM shape
RAMConfigurableSystem RAM depends on the selected cluster configuration; not published per GPU pricing row
GPU1× HGX B200180 GB VRAM / GPU
StorageConfigurableOptional managed shared storage (WEKA/VAST) can be attached to GPU clusters; capacity is configured separately
NetworkDedicated bare-metal cluster fabric · NVLink/NVSwitch inside node and high-speed InfiniBand for multi-node scale; exact per-offer port rate is configuration-specificprovider-reported network terms; port speed not reported
Available · updated 10:43 PM
Full specs
Together AIGPUDedicated inference on-demand

Together AI Dedicated Inference · HGX B200

Location not specified · Published / orderability not live-verified

$8.99/ hour≈ $6,562.70/mo
CPUConfigurableDedicated cluster-node CPU depends on the selected cluster configuration; Together publishes pricing per GPU, not per fixed VM shape
RAMConfigurableSystem RAM depends on the selected cluster configuration; not published per GPU pricing row
GPU1× HGX B200180 GB VRAM / GPU
StorageConfigurableOptional managed shared storage (WEKA/VAST) can be attached to GPU clusters; capacity is configured separately
NetworkDedicated bare-metal cluster fabric · NVLink/NVSwitch inside node and high-speed InfiniBand for multi-node scale; exact per-offer port rate is configuration-specificprovider-reported network terms; port speed not reported
Available · updated 10:43 PM
Full specs
Data source

What ComputeRadar reads from Together AI

Single-tenant dedicated inference GPU-hour pricing; it is deliberately separated from Together serverless token pricing.

The provider remains the seller. ComputeRadar normalizes API data for comparison and sends visitors to the provider for final configuration and checkout.

Official API documentation ↗

Provider FAQ

Together AI in ComputeRadar

Does ComputeRadar sell Together AI servers?

No. ComputeRadar is an independent comparison layer. The final order is made with Together AI.

Are prices guaranteed?

No. They reflect the latest API response available to ComputeRadar. Always confirm price, configuration, taxes and stock at provider checkout.

Can this page contain affiliate links?

Yes. When an approved partner tracking link is configured, outbound deal links can be affiliate links. This does not change the provider-reported price displayed by ComputeRadar.