SambaNova Cloud is an inference service where developers call open-source models through an API that follows the OpenAI chat completions format. After signing in you create an API key, store it in an environment variable and send requests to the api.sambanova.ai/v1 endpoint with curl, Python or the SambaNova SDK. The dashboard lists the available models and tracks your keys and usage. Streaming responses, system and user messages, and sampling controls such as temperature and top_p work the way existing OpenAI-style clients expect.
Software developers and AI engineers who want to build chat features, agents or reasoning workflows without hosting models themselves are the intended users. Teams already written against the OpenAI client libraries can usually switch by changing the base URL and key, which keeps migration effort small. Researchers comparing model families can try several from a single account, and hobbyists can use the Gradio helper to load a model into a quick demo.
Models on the dashboard include DeepSeek-V3.1 and DeepSeek-V3.2 for text and reasoning, Meta-Llama-3.3-70B-Instruct for text, MiniMax-M3 and gemma-4-31B-it for text and vision, and gpt-oss-120b for text. Typical projects include customer-facing chat assistants, coding helpers, document analysis and prototypes shared on Hugging Face through the Gradio deployment option. A developer can start with the provided curl example, then move to the Python client and wire the response into a web app. Switching between models means changing a single model name in the request, and the same key works for every model listed, which makes side-by-side comparisons quick for small teams and solo builders alike.
