IQ

Product Introduction

1. Definition: IQ Routing is a cloud-based AI gateway and management platform. Technically, it is a proxy layer and control plane that sits between a development team's applications and multiple Large Language Model (LLM) providers like OpenAI, Anthropic, and Google.
2. Core Value Proposition: It exists to provide centralized visibility, cost control, and operational efficiency for teams using multiple AI models. Its primary value is enabling AI spend management, usage monitoring, and intelligent model routing to reduce costs without locking developers into a single vendor.

Main Features

1. Unified Dashboard & Usage Analytics: The platform aggregates usage, spend, and performance metrics from all connected AI providers into a single dashboard. It works by intercepting API calls through its gateway (gateway.iq-routing.com), logging metadata (request counts, token usage, cost, latency, errors), and presenting it with team- and user-level breakdowns. It never stores the actual prompt or completion text.
2. Granular Access Control & Budgeting: It allows the creation of individual IQ API keys per application or team member, each with configurable RPM, TPM, and daily dollar limits. On the Team plan, it enforces daily and monthly budgets at the team level and provides an audit log of all administrative actions (key rotation, role changes).
3. Intelligent Model Routing & Cost Optimization: This is the core technical feature. The built-in routing engine can automatically direct requests to the most cost-effective model that meets the task's requirements. It works by allowing users to define "band maps" (e.g., use GPT-4 for complex synthesis, use Gemini Flash for simple retrieval) or by using a dynamic mode that selects models based on performance and cost. This decouples application logic from specific model APIs.
4. IQ Code Integration: A specific plugin for Claude Code that brings routing and cost visibility directly into the coding environment. It shows the measured cost and pace of AI-assisted coding sessions and allows routing decisions between models like Claude Opus and Haiku within the IDE.

Problems Solved

1. Pain Point: Uncontrolled and opaque AI spending. Teams lack visibility into which projects, users, or applications are driving costs across multiple provider accounts, leading to budget overruns.
2. Target Audience: Engineering leaders and platform teams at companies scaling their AI usage; Startup CTOs managing burn rate; Developers and DevOps engineers tasked with implementing and maintaining AI integrations.
3. Use Cases: Enforcing cost per feature: Assigning a dedicated IQ key with a strict daily budget to a new experimental feature. Onboarding and offboarding teammates: Quickly issuing or revoking individual access keys. Multi-model application architecture: Easily implementing a cost-optimized chain (e.g., using a cheaper model for classification and a powerful model for synthesis) without managing multiple client SDKs.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros: Exceptional focus on cost transparency and control through granular budgeting and key limits. The "bring your own keys" (BYOK) model avoids markup on model usage, aligning incentives. The routing feature provides tangible, measurable cost savings. The free tier is genuinely functional for small-scale testing.
   • Cons: Primarily a control plane; it does not replace a full-fledged LLM orchestration framework like LangChain for complex application logic. Advanced features like audit logs, team budgets, and custom routing require the paid Team plan. Savings from routing are workload-dependent and require configuration.

2. Key Alternatives & Differentiation:

   • OpenAI / Anthropic Native Dashboards: Provide basic usage data but only for their own models, with no cross-provider aggregation, budgeting, or routing. IQ centralizes control across a multi-vendor strategy.
   • Azure AI Studio / Google Vertex AI: Offer model management and deployment within their respective clouds but are primarily vendor-locked ecosystems. IQ is provider-agnostic, working with any OpenAI- or Anthropic-compatible endpoint.
   • Portkey, Martian Router: Similar AI gateway competitors. IQ differentiates with a stronger emphasis on the team-level financial controls (budgets, per-key dollar caps) and its unique IDE integration via IQ Code, catering to both application and developer workflow costs.

Frequently Asked Questions (FAQ)

1. How does IQ Routing save money on AI API costs? IQ Routing saves money through intelligent model routing, directing each API request to the most cost-effective model capable of handling the task. It also provides budgets, alerts, and per-key spending caps to prevent runaway costs before they happen.

2. Does IQ Routing store or read my prompt data? No. According to its documentation, IQ Routing logs metadata for analytics—such as request counts, token usage, and cost—but never stores or has access to the actual text of your prompts, completions, or code.

3. Can I use IQ Routing with my existing OpenAI or Anthropic API code? Yes. Integration is designed for minimal code change. You typically only need to replace your provider's base URL with the IQ gateway URL (e.g., base_url="https://gateway.iq-routing.com/v1") and use an IQ API key. Your existing OpenAI SDK or Anthropic SDK calls will work.

4. What's the difference between the Free and Team plans? The Free plan includes basic gateway access, per-key rate/limit controls, and support for your own provider keys. The Team plan ($70/org/month) adds team-level daily/monthly budgets, spend/error/latency alerts, an audit log, member roles (Owner, Admin, Member), and advanced routing modes.

5. How does IQ Code relate to the main IQ Routing dashboard? IQ Code is a plugin for Claude Code that uses the IQ Routing backend. It allows individual developers to apply routing and cost tracking within their IDE. Usage from IQ Code can be reported to the main team dashboard if the "Share usage facts" setting is enabled, providing visibility into development costs.