Pricing · 2026
Together AI Dedicated Inference pricing
3 plans on the record
Every plan
Dedicated Inference — NVIDIA HGX H100 (on-demand)
Priced on request
On-demand, pay as you go; billed per GPU per hour
Single-tenant GPU instances · guaranteed performance (no sharing) · custom models · autoscaling and traffic spike handling
Read 8 Oct 2026 · together.ai
Dedicated Inference — NVIDIA HGX B200 (on-demand)
Priced on request
On-demand, pay as you go; billed per GPU per hour
Single-tenant GPU instances · guaranteed performance (no sharing) · custom models · autoscaling and traffic spike handling
Read 8 Oct 2026 · together.ai
Dedicated Inference — reserved capacity
Priced on request
Contact sales
Reserved GPU capacity · pricing depends on hardware and configuration
Read 8 Oct 2026 · together.ai