Model providers
LLM providers for AI agents
Connect a provider once. Its key stays in Dome, and every agent call to it is authorized, metered and audited.
Model providers
LLM providers for AI agents
Any OpenAI-compatible provider
Plus Anthropic, Amazon Bedrock, Vertex AI and Azure OpenAI natively.
AI21Providers
Amazon BedrockProvidersAmazon Bedrock behind the Model Broker: AWS credentials held in Dome, Converse calls signed per request, Claude pooled with Anthropic.
AnthropicProvidersThe Claude API behind the Model Broker: the key held in Dome, every call authorized, metered and audited.
Azure OpenAIProvidersAzure OpenAI deployments behind the Model Broker: the key held in Dome, region rules on who reaches which deployment.
CerebrasProvidersThe Cerebras API behind the Model Broker: open models with the key held in Dome, Fireworks for failover and full audit.
CohereProvidersCohere's Command and Embed models behind the Model Broker: the key held in Dome, approval rules per team, every call audited.
DatabricksProvidersDatabricks Model Serving behind the Model Broker: the token held in Dome, a vendor fallback in the same pool, every call audited.
DeepInfraProvidersThe DeepInfra API behind the Model Broker: open models with the key held in Dome, Together AI for failover and full audit.
DeepSeekProvidersThe DeepSeek API behind the Model Broker: the key held in Dome, residency rules on regulated work, every call audited.
Fireworks AIProvidersThe Fireworks AI API behind the Model Broker: open models with the key held in Dome, Together AI for failover and full audit.
Google GeminiProvidersGemini on Vertex AI behind the Model Broker: the service account held in Dome, every call authorized, metered and audited.
GroqProvidersThe Groq API behind the Model Broker: open-weight models with the key held in Dome, a fallback provider and full audit.
MistralProvidersThe Mistral API behind the Model Broker: the key held in Dome, region rules on EU work, every call audited.
Moonshot AIProvidersThe Moonshot AI Kimi API behind the Model Broker: the key held in Dome, trials limited to one team, every call audited.
NVIDIA NIMProvidersNVIDIA NIM behind the Model Broker: self-hosted containers first, hosted NIM as failover, and rules on who may use the hosted API.How it works
Every provider works the same way
Add a connection with the provider's key, then point agents at the Gateway. OpenAI- and Anthropic-compatible clients both work.
01
Add the connection
The key goes into Dome's vault, per workspace and connection, and is injected at call time.
$ dome models add claude-fable \--provider anthropic --model claude-fable-5-1 \--api-key "$ANTHROPIC_API_KEY" --gateway prod-gateway02
Point agents at the Gateway
The agent's Dome key replaces the provider key.
from openai import OpenAIclient = OpenAI(base_url=f"{GATEWAY_URL}/v1", api_key=DOME_AGENT_KEY)
Commands tested against a Dome workspace on September 30, 2026.
FAQ
Common questions
Where are provider keys stored?
In Dome's vault, per workspace and connection, and injected at call time. Agents never hold them.
Can I use a provider that isn't listed?
Yes, if it serves the OpenAI-compatible API. Add it as an OpenAI-compatible connection with its base URL.
Can one agent use several providers?
Yes. Pool connections from different providers, and the agent sends one model name.
Next steps
Talk with our FDE team
Our forward deployed engineers work with your platform team to get your agents into production and under control: the first one governed on your own systems, and a pattern your teams can repeat for every agent after it.
No card required to start. Register your first agent in minutes.
LLM router guide
What an LLM router does, and what to ask of one before your agents depend on it.
Read the guide