AI Infrastructure Pricing
Summary
Together AI offers transparent pricing for AI infrastructure, including token-based serverless inference and dedicated GPU instances. Rates vary by model and hardware, with discounts for reserved capacity. Additional services cover code sandboxes, interpreters, and model fine-tuning.
Highlights
- Token-based serverless inference pricing varies by model, with cached output rates often lower.
- Dedicated single-tenant GPU instances are available for hardware like H100 and HGX B200.
- Reserved GPU capacity offers discounted hourly rates for commitments from 7 to 180+ days.
- Code sandboxes and interpreters are priced per hour or session, with filesystem storage fees.
- Supervised fine-tuning is priced per 1M tokens, with tiers for standard and specialized needs.
auto-generated
Context
Audience
AI Engineers and Data Scientists
DomainMachine Learning Infrastructure
Formatpricing page
Accessfree online
Topics
Related
LLM API PricingGPU Cloud ComputingServerless InferenceModel Fine-TuningNVIDIA H100NVIDIA HGX B200