Providers /SambaNova
Inference provider · HQ US

SambaNova

Models
12
Regions
00

Our take

Editorial by LLMap · updated Jul 28, 2026

A US-based, enterprise-grade host with verified SOC 2 compliance and a small, curated catalogue of 8 offers. It delivers strong throughput on select models but charges mid-market to premium prices with narrow coverage.

Use this for enterprise workloads that require verified SOC 2 and a guarantee that prompts are not used for training. Pick it when throughput of 104.5–165 tokens per second meets your service-level needs, or when a specific model in its 8-offer catalogue is required and alternatives lack it. Skip it if you need broad model choice, EU data residency, or the lowest price for a given model.

Strengths
  • Verified enterprise compliance posture — SOC 2 is confirmed and it does not train on prompts.
  • Strong throughput on select models: MiniMax M2-7 at 165 tokens per second (the highest we have tracked for that model) and Gemma 4 31B at 104.5 tokens per second (the second-highest we have tracked).
  • Tiered pricing options for flexibility — two price and throughput tiers for MiniMax M2-7 and DeepSeek V3.1 Terminus.
Trade-offs
  • Small catalogue limits model choice: only 8 tracked offers, compared with 61 at volume leaders.
  • Premium pricing on latest and reasoning models — DeepSeek V3.2 at $3/$4.5 is among the highest tracked prices for that model family.
  • Throughput is unmeasured for the majority of offers: 5 of 8 have no throughput data, including flagship-class models such as GPT-OSS 120B, Llama 3.3 70B and DeepSeek V3.2.
01

Privacy & data handling

Yes
No
No answer on record

The colour says whether the answer favours you, not whether it is a yes — not training on your prompts earns a green cross, no zero-retention option earns a red one.

Trains on your prompts
No
Logs prompts
No
Offered
Attested

Prompt retention: none.

02

Models & pricing

Something wrong on this page? Tell us