Install Jentic One Beta
Jentic One is a self-hosted execution layer for AI agents. It lets your agent call the SambaNova Cloud API, or any other public or private API you need. You set the rules, the agent never sees your credentials, and every call is logged.
Two steps, two machines. Install the instance in a safe environment, then register your agent from wherever it runs.
Step 1: Jentic One Host machine
# On the machine that will host your Jentic One instance:
curl -fsSL "https://jentic.com/install.sh?src=apis&api=%2Fapis%2Fsambanova.ai%2Fsambanova" | shStep 2: Agent machine
# On the machine where your agent runs (keep this separate from the instance):
curl -fsSL "https://jentic.com/install.sh?src=apis&api=%2Fapis%2Fsambanova.ai%2Fsambanova" | sh
jentic register # connects your agent to your Jentic One instanceJentic One is in public beta. The setup above keeps your agent separate from the instance, which is what you want before using real credentials: an agent running as the same OS user as Jentic One can read its stored keys directly. Just evaluating? A single local install is fine to start. See the secure deployment guide for the tiers.
What an agent can do with SambaNova Cloud API.
Generate a chat completion from a model and a messages list, with optional streaming
Tune sampling with temperature, top_p, top_k, repetition_penalty, and max_tokens
Request structured output through the response_format field on chat completions
Produce a plain text completion from a single prompt
Create vector embeddings from input text for semantic search
GET STARTED
List the models available to your SambaNova Cloud key before choosing one
Patterns agents use SambaNova Cloud API for, with concrete tasks.
★ AI agent using an open model backend
A developer wants their agent to call an open model for a sub-task, such as drafting or classification, without wiring a second SDK. Through Jentic the agent searches for 'create a chat completion with Llama', loads createChatCompletion with its input schema, and sends the required model and messages fields while the SambaNova key is injected at call time. The response arrives in the familiar OpenAI chat-completion shape.
POST /chat/completions with model Meta-Llama-3.1-405B-Instruct and a two-message conversation, then return the assistant reply
Drop-in OpenAI replacement
Applications written against OpenAI-style chat and completion requests can point at SambaNova Cloud to run open models. The request bodies use the same model, messages, prompt, max_tokens, temperature, stream, and stop fields, so switching is mostly a base URL and key change, with GET /models confirming which model ids exist.
Call GET /models, pick a DeepSeek model id from the list, and send the same prompt via POST /completions
Embeddings for retrieval
Retrieval-augmented generation needs vectors for documents and queries. POST /embeddings takes a model and input and returns embeddings that can be stored in a vector database and searched before calling chat completions with the retrieved context.
Embed five product descriptions with POST /embeddings and return the vectors for indexing
4 endpoints — jentic publishes the only available openapi specification for sambanova cloud api, keeping it validated and agent-ready.
METHOD
PATH
DESCRIPTION
/chat/completions
Create a chat completion
/completions
Create a text completion
/embeddings
Create embeddings
/models
List models
/chat/completions
Create a chat completion
/completions
Create a text completion
/embeddings
Create embeddings
/models
List models
What agents get from Jentic-routed access to this vendor.
Setup
Wiring SambaNova by hand means pointing an OpenAI-style client at api.sambanova.ai/v1, attaching the bearer key, and checking model ids against the models list. Through Jentic you install once, import SambaNova Cloud from the API Directory, store the key once, and your agent calls it.
Permission scoping
SambaNova selects the model in the request body, so rules bound which operations your agent may call, such as embeddings only, not which model it uses. You choose the operations it may call, so chat and text completions are not included unless you add them.
Credential isolation
Your SambaNova Cloud API key is stored once, encrypted, by your own Jentic One instance and injected at execution time. It never enters the agent's prompt, logs, or context.
Intent-based discovery
Agents search Jentic by intent such as 'create a chat completion' or 'embed text', and Jentic returns the matching SambaNova operation with its input schema so the agent sends the right model and message fields.
Alternatives and complements available in the Jentic catalogue.
Specific to using SambaNova Cloud API through Jentic.
Why is there no official OpenAPI spec for SambaNova Cloud API?
SambaNova does not publish an OpenAPI specification. Jentic generates and maintains this spec so that AI agents and developers can call SambaNova Cloud API via structured tooling. It is validated against the live API and kept up to date. To run it on your own infrastructure, install Jentic One from its GitHub repo.
What authentication does the SambaNova Cloud API use?
Per its OpenAPI spec, the SambaNova Cloud API authenticates with your SambaNova Cloud API key sent as a bearer token in the Authorization header. Through Jentic the key is stored by your own instance and attached at call time, so the agent never handles it.
Is the SambaNova Cloud API compatible with OpenAI clients?
Yes. The spec describes an OpenAI-compatible API: POST /chat/completions takes model and messages, POST /completions takes model and prompt, and POST /embeddings takes model and input, matching the OpenAI request shapes at the base URL https://api.sambanova.ai/v1.
What are the rate limits for the SambaNova Cloud API?
The OpenAPI spec does not specify rate limits. Check the SambaNova Cloud documentation at https://cloud.sambanova.ai/docs for current per-model limits.
Is there a SambaNova MCP server?
You don't need an MCP server to give your agent SambaNova Cloud. Jentic connects it directly from the API Directory: import it, store your API key once, and your agent calls the chat, completion, and embedding operations when it needs them.
How do I get a chat completion from SambaNova through Jentic?
Search Jentic for 'create a chat completion with Llama', load createChatCompletion, and execute POST /chat/completions with a model id from GET /models and a messages array. Set stream to true for token-by-token output.
Can I limit what my agent is allowed to do with the SambaNova Cloud API?
Yes. For a retrieval pipeline, write a rule that allows only POST /embeddings and GET /models, so the agent can index documents but cannot run chat or text completions on your key. Each call it makes is logged.
Know of an official OpenAPI document? Contribute it →
For Agents
Generate chat or text completions and embeddings with open models like Llama and DeepSeek on SambaNova Cloud, using OpenAI-style requests.
Use for: I need a fast Llama chat completion, Summarise this document with DeepSeek-R1, Stream a long answer token by token, Get embeddings for these product descriptions
Not supported: Does not fine-tune models, generate images, or transcribe audio. Use for chat completions, text completions, and embeddings on SambaNova Cloud only.
Jentic publishes the only available OpenAPI specification for SambaNova Cloud API, keeping it validated and agent-ready. The SambaNova Cloud API serves open large language models such as Llama, DeepSeek, and QwQ on SambaNova's custom AI chips through an OpenAI-compatible interface. It generates chat completions from a list of messages, plain text completions from a prompt, and vector embeddings from input text, and lists the models available to your key. Chat requests support streaming, stop sequences, top_p and top_k sampling, repetition_penalty, and a response_format option.
This API is usable in Jentic One now. Its AI-readiness score against Jentic's framework shows where it stands today and where improvements would make it even easier for agents to use.
Base layer of spec validity and structural soundness.
Aggregated quality score from linter diagnostics, weighted by severity.
Percentage of `$ref` references that resolve successfully.
Checks whether the API description parses successfully and conforms to its declared specification (e.g., OpenAPI).
Structural correctness score based on schema issues using logarithmic dampening.
Clarity, completeness, and ingestion readiness for developers and tooling.
How richly the API is illustrated with examples.
Percentage of examples that conform to their schemas.
Percentage of operations with complete response definitions (success, client error, server error).
Health of API ingestion, bundling, and resolution within Jentic pipelines.
Semantic breadth, depth, and agent comprehension for AI systems.
Coverage of descriptions across API elements.
Coverage of RFC 9457 Problem Details for error responses.
Coverage, uniqueness, and casing consistency of operationIds for AI inference.
Coverage of summaries across operations/tags/info.
Functional utility, complexity comfort, and AI orchestration readiness.
Agent comfort level based on API operational and structural complexity.
Trust, risk posture, and security compliance.
Average quality of security schemes based on authentication method strength (weakest link for OAuth2).
Findability, semantic richness, and reasoning readiness.
Clarity and depth of descriptions across API elements.
Score it yourself
Every API in the directory is allowlisted, so you can re-score it with no key required.
npx @jentic/api-scorecard-cli score <openapi-url>