Cerebras API and Dome
Fast inference on Cerebras, with every call governed. Cerebras is the provider you reach for when a person is waiting on the answer. Connect it to Dome once and its capacity goes to those agents: a rule keeps batch work off it, and Fireworks AI takes gpt-oss-120b calls when Cerebras fails.
Dome provides model brokeringand routing for Cerebras.
One Cerebras key, many agents
Dome holds the Cerebras key and sends it as a bearer token. Each voice and routing agent calls with its own Dome key, and audit names it.
Same model, second provider
Fireworks AI serves gpt-oss-120b too. Pool the two and a failed call on Cerebras goes to Fireworks.
Fast capacity where it counts
Tag Cerebras connections as the fast lane, and a rule keeps batch agents off it.
Connect Cerebrasin three steps.
Add Cerebras, back it with Fireworks AI, then point agents at the Gateway.
Add the connection
The provider id is cerebras. Dome uses https://api.cerebras.ai/v1 and Cerebras' own model ids.
shelldome models add gpt-oss-cerebras \ --provider cerebras \ --model gpt-oss-120b \ --api-key "$CEREBRAS_API_KEY" \ --attributes '{"lane":"fast"}' \ --gateway prod-gatewayPool it with Fireworks AI
Add accounts/fireworks/models/gpt-oss-120b as a second connection. Members are tried in priority order.
shelldome models pool create gpt-oss-120b \ --failover-max all --gateway prod-gateway dome models pool member add gpt-oss-120b gpt-oss-cerebras --priority 0dome models pool member add gpt-oss-120b gpt-oss-fireworks --priority 1Point your agent at the Gateway
Any OpenAI-compatible client works. The agent's Dome key replaces the Cerebras key.
pythonfrom openai import OpenAI client = OpenAI(base_url=f"{GATEWAY_URL}/v1", api_key=DOME_AGENT_KEY)client.chat.completions.create( model="gpt-oss-120b", messages=[{"role": "user", "content": "Classify this caller's intent."}],)
Keep batch agentsoff the fast lane.
Applied to a batch agent with --agent, this refuses any connection tagged lane: fast. Cerebras capacity goes to the agents a person is waiting on.
cedar
forbid (principal, action == Dome::Action::"llm:invoke", resource is Dome::LLMModel)when { resource has lane && resource.lane == "fast" };Bringing it alltogether.
Connecting Cerebras to registered agents, tools, and identity in Dome completes a governed agent application.
Acting for
Control point
GatewayEvery call decided and audited
RulesGuardsQuotas
Models
Common questionsabout Cerebras.
More of whatDome works with.
Model
Claude Fable
Fable 5.1 from Anthropic and Amazon Bedrock in one failover pool, open to one group and capped by quota.
MCP server
GitHub
The GitHub MCP server behind the Tool Gateway, with rules that decide each call on its owner and repository.
Identity
Okta
Okta tokens verified on every agent call, so rules and audit name the person each agent acted for.
Runtime
LangGraph
LangGraph agents with their model calls on the Model Broker and their MCP tools on the Tool Gateway.
Client
Claude Code
Claude Code on a Dome Gateway with per-developer sign-in, rules on every tool call, and audit by name.
Agent service
TinyFish
TinyFish's web agents behind the Tool Gateway, with rules that decide each run on the site it targets.
Talk with ourFDE team.
Or start free, and register your first agent in minutes. Connect it through a gateway, give it rules, and see every call it makes in one audit trail.
Talk with our FDE team
Our engineers design, build and run governed agents for the part of your business you want to change. You see what they deliver.
LLM providers for AI agents: one governed path to every model
Connect a provider once. Its key stays in Dome, and every agent call to it is authorized, metered and audited.