DeployWhere

Together AI

AI cloud platform for serverless and dedicated inference, GPU compute, storage, sandboxes, and fine-tuning.

No reviews yet. Write the first.
Visit Together AI

Together AI is a GPU and AI hosting provider based in United States. No European datacenter locations are recorded for it yet. A free tier is available. As a US company it falls under the US CLOUD Act. Pricing on this page is published only after a human has reviewed it.

PricingPros and consSpecificationsAboutReviews
Company based in
United States
European regions
None listed
US CLOUD Act
US company
GDPR agreement
Available
Free tier
New users receive $25 in free credits.
VAT
Shown excluding VAT
Together AI website
Website image published by Together AI.

Pricing

9 plans
  • Serverless Inference
    No provisioning or minimum cost. Models are charged per 1M tokens, images per megapixel or image, video per video, audio per 1M characters or audio minute. New users get $25 in free credits.
    Monthly
    Not listed
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed
  • Provisioned Throughput
    Price is $0.05 per PTU/MIN for listed models. PTU capacity is reserved and the page states a 99% uptime SLA.
    Monthly
    $0.05
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed
  • NVIDIA HGX H100
    Dedicated Inference, on-demand price per GPU per hour. Reserved pricing requires contacting sales.
    Monthly
    $5.49
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed
  • NVIDIA HGX B200
    Dedicated Inference, on-demand price per GPU per hour.
    Monthly
    $8.99
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed
  • GPU Clusters
    All prices are per GPU per hour. On-demand rates include NVIDIA HGX H100 $3.99, NVIDIA HGX H200 $5.99, and NVIDIA HGX B200 $8.19. Reserved rates are listed for 7-30, 31-90, 91-180, and 181+ days, with longer reservations discounted. Some hardware shows no published rate and requires contacting sales.
    Monthly
    Not listed
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed
  • Code Sandbox
    Per vCPU is $0.0446 per hour and per GiB RAM is $0.0149 per hour.
    Monthly
    Not listed
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed
  • Code Interpreter
    Price is per 60-minute session.
    Monthly
    $0.03
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed
  • Shared Filesystem
    Price is $0.16 per GiB/month. The homepage states zero egress fees for Managed Storage.
    Monthly
    $0.16
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed
  • Fine-Tuning
    Standard pricing is per 1M tokens. Supervised Fine-Tuning ranges from $0.48 to $3.20 and Direct Preference Optimization from $1.20 to $8.00, depending on model size and method. Each job has a minimum charge of $4.00. Specialized model pricing is also listed.
    Monthly
    Not listed
    vCPU
    Not listed
    Memory
    Not listed
    Storage
    Not listed

Pros and cons

What works

  • Offers serverless inference without provisioning or minimum cost.
  • Provides both usage-based inference and dedicated GPU capacity.
  • Lists reserved GPU cluster rates for multiple commitment periods.
  • Managed Storage is stated to have zero egress fees.

What to watch

  • Many dedicated hardware and reserved prices require contacting sales.
  • Serverless pricing varies by model and modality rather than using a single platform rate.
  • No datacenter locations, data residency guarantee, or compliance certifications are stated.

Specifications

Egress policy
The homepage states zero egress fees for Managed Storage; no general egress policy is stated.
Uptime sla
99% for Provisioned Throughput
Support
The pages provide contact-sales links but do not state support channels or hours.

About Together AI

Together AI provides serverless and dedicated model inference, GPU clusters, code sandboxes, managed storage, and fine-tuning. Serverless inference has no provisioning or minimum cost and charges by processed usage, while dedicated inference and GPU clusters charge per GPU hour. The pages do not state the operating company's domicile or datacenter locations.

Reviews

None yet

No approved reviews yet. Every review is read by a person before it appears, and only approved reviews count toward a rating.

Other gpu and ai hosting

See all
DataCrunch
Finland
Exoscale GPU
Switzerland
IONOS AI
Germany
Nebius
Netherlands
OVHcloud AI
France
Scaleway GPU Instances
France

Something here out of date?

Every figure on this page was checked by a person, and providers change their plans without telling anybody. Say what moved and it gets checked again.

Work at Together AI? Claim this listing to correct the description and add screenshots. Verification is a confirmation link to an address on your own domain. Pricing still goes through the same check as every other listing here.