Together AI GPU Pricing, Specs & Review
Last updated: 2026-03-17
Overview
Together AI is a full-stack AI platform offering inference APIs, fine-tuning, and dedicated GPU clusters. Built on cutting-edge research, Together AI provides instant access to NVIDIA B200, H200, H100, and L40S GPU clusters with flexible hourly and reserved pricing. Their platform is optimized for both inference and training workloads, with competitive pricing starting at $2.10/hr for L40S and $2.69/hr for H100 instant clusters.
GPU Pricing
| GPU | VRAM | On-Demand $/hr | 1-6 Days $/hr | 1 Week+ $/hr |
|---|---|---|---|---|
| B200 | 192 GB | $5.49 | $4.79 | Contact |
| H200 | 141 GB | $3.19 | $2.79 | $3.15 |
| H100 SXM | 80 GB | $2.69 | $2.39 | $2.20 |
| L40S | 48 GB | $2.10 | N/A | N/A |
| A100 PCIe | 40 GB | $2.40 | N/A | N/A |
Key Features
- Full-stack AI platform: inference, fine-tuning, and GPU clusters
- NVIDIA B200, H200, H100, L40S, and A100 GPUs
- Instant GPU clusters available in minutes
- Reserved clusters with discounted weekly+ rates
- Serverless inference API for 100+ open-source models
- Fine-tuning as a service with pay-per-token pricing
- Research-backed optimization for training and inference
- Up to 64 GPUs per instant cluster
Pros & Cons
Pros
- Competitive H100 pricing ($2.69/hr instant, $2.20/hr reserved)
- Full-stack platform combining inference, fine-tuning, and clusters
- Instant cluster provisioning in minutes
- Strong open-source model ecosystem
- Research-backed performance optimizations
Cons
- Smaller GPU selection compared to hyperscalers
- No spot/preemptible pricing option
- Fewer regions and data center locations
- Reserved pricing requires contacting sales for longest terms
FAQ
What GPU models does Together AI offer?
Together AI offers NVIDIA B200, H200, H100 SXM, L40S, and A100 PCIe GPU instances through instant and reserved cluster deployments.
What is the cheapest GPU option on Together AI?
Together AI L40S instances start at $2.10/hr. H100 SXM instant clusters are available from $2.69/hr, with reserved rates as low as $2.20/hr for weekly+ commitments.
Does Together AI offer reserved GPU pricing?
Yes. Together AI offers tiered pricing: instant hourly rates, discounted 1-6 day rates, and further discounts for 1-week+ reservations. Contact sales for longer-term custom pricing.