LaunchRoundsSubmit Software

Orq: One gateway for models, agents and spend

Orq is an AI gateway that routes every model, agent and tool request through a single control point where access, cost and compliance policies are enforced.

What is Orq?

Orq places a gateway in front of every AI request so that model calls and tool calls pass through one point before a response is returned. From there it handles routing across 500+ models from 30+ providers, fallbacks, retries, caching and load balancing, and it applies policies such as PII redaction, budget caps and failover rules at request time. Traces, token usage and cost are attributed by user, team, project, customer or agent, and the same controls extend to tool access through an MCP gateway. The platform also covers evaluation and observability, with offline and live testing against a curated dataset and an LLM-as-a-judge setup, plus a knowledge base for RAG workflows and an agent runtime. It is framework agnostic, working with OpenAI Agents, Vercel AI, CrewAI, LangGraph and OpenTelemetry stacks. Deployment options range from managed EU hosting to the customer's own cloud, on-premises and fully air-gapped environments.

Focus areas: AI Gateway, Guardrails.

Who is Orq for?

  • Platform and infrastructure teams running many AI agents
  • Security and compliance teams needing audit trails and access limits
  • AI engineering teams managing model spend across providers

Is Orq worth it?

A practical control layer for teams that need routing, spend caps and audit trails across many models and agents

Where Orq is strong

  • Single gateway for routing, guardrails and cost attribution
  • Framework agnostic with OpenTelemetry support
  • EU hosting plus on-premises and air-gapped deployment options

Where it falls short

  • No public pricing, so cost must be confirmed with sales
  • Published outcome figures on the site appear as placeholders

How do you get started with Orq?

  1. 01Get an API key and route requests through the gateway
  2. 02Connect model providers and set routing and fallback rules
  3. 03Define budget caps and access policies per team or agent

Orq: frequently asked questions

What does Orq actually do?

It routes every AI request through one gateway, where it applies routing rules, PII redaction, budget caps and failovers, and records traces, token usage and cost.

Which models and frameworks does it support?

The site states one API key covers 500+ models across 30+ providers, and that it works with OpenAI Agents, Vercel AI, CrewAI, LangGraph and any OpenTelemetry stack.

Can I control what agents are allowed to access?

Yes. Access control is applied at the tool level through the MCP gateway, with scoped tools per team and an audit log of which agent called which tool.

Where is it hosted?

Managed hosting is in the EU by default, with options for the customer's own cloud, on-premises deployment and fully air-gapped setups.

Orq alternatives

The closest tools in our directory, matched on category, focus areas and audience.

Orq compared with its alternatives
ToolBest forFree planStarting priceFocus areas
OrqTeams governing many AI agents, models and tool callsUnknown—AI Gateway, Guardrails
PortkeyTeams taking LLM apps to production with monitoring and controlYes$49/moAI Gateway, Guardrails, Prompt Management
APIParkTeams needing one open-source gateway for AI and API trafficUnknown—AI Gateway
TrueFoundryTeams centralizing LLM, MCP and agent traffic with governanceYes$25/moAI Gateway, Model Deployment
HigressTeams centralizing LLM, Agent and MCP traffic behind one gatewayUnknown—AI Gateway, MCP
Mistral AITeams needing controllable, self-hostable language modelsYes$14.99/mo—
Upstage AITeams automating document parsing and extraction with LLM APIsNo$100/mo—
PromptfooDevelopers and security teams testing AI apps for vulnerabilitiesYes—Guardrails
CohereEnterprises needing private, controllable generative AI and retrievalNo——
Maxim AITeams centralising LLM access, cost control and observabilityYes—AI Gateway
Not DiamondTeams cutting LLM inference cost on coding agentsNo——