What is SambaNova Turbo?

SambaNova Cloud provides high-speed hosted inference with free evaluation, pay-as-you-go developer access and enterprise capacity.

Core offerings

  • Production and preview language models.
  • API keys, playground and usage metering.
  • Cached-input pricing on supported models.
  • Enterprise subscriptions, custom limits and deployment options.

Pricing

MiniMax-M2.7 and MiniMax-M3 each cost US$0.60 input, US$0.06 cached input and US$2.40 output per million tokens. DeepSeek V3.1 and V3.2 cost US$3 input and US$4.50 output. Gemma 4 31B costs US$0.38/US$1.15, GPT OSS 120B US$0.22/US$0.59 and Llama 3.3 70B US$0.60/US$1.20.

API and model availability

All seven non-retired models on SambaNova's current official table are reproduced on the API Cost page.

Official sources: SambaNova plans, model pricing.