Hosted Service
Inference & Serving
SambaNova Cloud
1Custom 2RDU 3chips 4power 5rapid 6inference
the six Ws · specification
W1
Who
Teams wanting high-throughput inference on open models without owning specialized hardware.
W2
What
SambaNova Cloud is a hosted inference API running Llama, DeepSeek, and other open models on SambaNova's custom RDU chips.
W3
Where
Accessed through SambaNova's cloud API endpoint.
W4
When
Public cloud launched in 2024 with a developer tier added in 2025 and ongoing 2026 model updates.
W5
Why
It exists to offer fast, cost-competitive inference as an alternative to GPU-based providers.
W6
With
Provides an OpenAI-compatible API usable with standard LLM client SDKs.
W7
Watch
SaaS, proprietary, requires an API key; free, developer, and enterprise tiers with free trial credit; data processed on SambaNova's cloud.
alternatives
works with
RDU chipsfast inferencepay-per-tokenopen models