Announcing Dome Managed Agents.

Try it now

Qwen and Dome

Run Qwen 3.8 from two providers under one name. Teams pick Qwen for coding agents that run all day on open weights. The governance question is scope: a coding agent pinned to Qwen shouldn't wander onto a frontier model on a bad prompt. Dome pins it, backs Together AI with DeepInfra, and budgets each agent.

Dome provides model brokeringand routing for Qwen.

Together first, DeepInfra next

Both serve Qwen/Qwen3.8-2.4T-A95B. Pool them as qwen3.8, and a failed call moves to the next provider.

A coding agent that stays on Qwen

Tag Qwen connections with a family attribute. A rule on the agent refuses every model outside it.

Spend by agent

Cap each agent's monthly provider spend. The migration agent can't spend the test writer's budget.

Qwen with failoverin five steps.

Together and DeepInfra under qwen3.8, a pin on the coding agent, and a budget per agent.
  1. Connect Together AI

    Tag it family: qwen. The Together key is Dome's to hold; the coding agent gets a Dome key.

    shell
    dome models add qwen-together \  --provider together \  --model Qwen/Qwen3.8-2.4T-A95B \  --api-key "$TOGETHER_API_KEY" \  --attributes '{"family":"qwen"}' \  --gateway prod-gateway
  2. Connect DeepInfra

    DeepInfra serves the same weights under the same id.

    shell
    dome models add qwen-deepinfra \  --provider deepinfra \  --model Qwen/Qwen3.8-2.4T-A95B \  --api-key "$DEEPINFRA_API_KEY" \  --attributes '{"family":"qwen"}' \  --gateway prod-gateway
  3. Pool them for failover

    Same id on both hosts, so the pool is a straight priority list: Together, then DeepInfra.

    shell
    dome models pool create qwen3.8 \  --failover-max all --gateway prod-gateway
    dome models pool member add qwen3.8 qwen-together --priority 0dome models pool member add qwen3.8 qwen-deepinfra --priority 1
  4. Point your agent at the Gateway

    The code assistant's OpenAI client sends qwen3.8 to the Gateway with its own key.

    python
    from openai import OpenAI
    client = OpenAI(base_url=f"{GATEWAY_URL}/v1", api_key=DOME_AGENT_KEY)client.chat.completions.create(    model="qwen3.8",    messages=[{"role": "user", "content": "Write unit tests for parse_invoice()."}],)
  5. Cap spend per agent

    Each agent's budget follows it across pools, so a second model doesn't open a second budget.

    shell
    dome quotas set --subject agent --agent migration-agent \  --dimension llm --unit provider_usd --limit 150 --window monthly

Keep an agenton Qwen.

Applied to one agent with --agent, this refuses any connection not tagged as Qwen. The code assistant can't reach Opus, whatever model name it sends.
cedar
forbid (principal, action == Dome::Action::"llm:invoke", resource is Dome::LLMModel)unless { resource has family && resource.family == "qwen" };

One call,two outcomes.

Switch the caller or the argument and watch the same call decide differently, with every decision landing in audit.

Model

agent code-assistant · acting as d.chen
llm:invoke(model: "qwen3.8")
  1. Agentcode-assistant is registered and active
  2. Callerd.chen verified
  3. RuleConnection is tagged family: qwen
DecisionAllowed

Bringing it alltogether.

Connecting Qwen to registered agents, tools, and identity in Dome completes a governed agent application.
Dome

Control point

GatewayEvery call decided and audited
RulesGuardsQuotas

Common questionsabout Qwen.

Which providers serve Qwen through Dome?
Together AI and DeepInfra serve Qwen 3.8 as Qwen/Qwen3.8-2.4T-A95B. Fireworks AI, Cerebras and Groq serve other Qwen models under their own ids.
Can the agent get around the rule by asking for another model?
No. The rule reads the tag on the connection that would serve the call, not the name the agent sends.
What happens when a new Qwen ships?
Connect the new Qwen on Together and DeepInfra with the same family tag. The pin still holds, and the pool decides which version agents get.

More of whatDome works with.

Provider

Anthropic

The Claude API behind the Model Broker: the key held in Dome, every call authorized, metered and audited.
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

OpenAI Agents SDK

OpenAI Agents SDK agents with their models 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.

AI models for enterprise agents: failover, rules and quotas

Every model your agents call goes through the Model Broker. Pool providers for failover, decide who may call each model, and cap what it costs.

Every agent you run, under one system of control.