Zerg Router

Product Introduction

1. Definition: Zerg Router, also branded ZergRouter, is an OpenAI-compatible API routing gateway and model access management layer for AI coding agents. It sits between supported developer tools such as Codex, OpenCode, Pi, ZTC, and other OpenAI-compatible applications, and model providers such as Modal. It exposes a unified API endpoint at https://zergrouter.com/v1, including routes such as /v1/responses and /v1/models. It is part of the ZergAI ecosystem at https://zergai.com and is classified as a DeveloperApplication and SoftwareApplication. Zerg Router centralizes routing policy, provider credentials, scoped Router keys, per-key budgets, fallback chains, Codex quota monitoring, and routed usage analytics.
2. Core Value Proposition: Zerg Router exists so developers can keep their existing coding tools while choosing models and enforcing account-level routing policy. Instead of configuring model selection, API limits, fallbacks, and provider keys separately inside every tool, users define these policies once in their Zerg account. The product provides one OpenAI-compatible endpoint for coding agents, per-API-key daily budgets enforced per request, explicit fallbacks on 5xx, 429, or timeout, and visibility into remaining weekly Codex quotas, reset dates, and banked resets. It supports paid platform usage or bring-your-own provider keys.

Main Features

1. OpenAI-Compatible Unified Router Endpoint: Zerg Router exposes an OpenAI-compatible API at https://zergrouter.com/v1, with a primary example route at /v1/responses. Supported clients such as Codex, Pi, OpenCode, ZTC, and OpenAI-compatible apps connect through a Router key and send requests to one endpoint. The router authenticates the request, checks account-level routing policy, selects the allowed model route, and forwards the request to the configured provider or model. The endpoint is designed for supported chat requests and Codex Responses workflows, and model availability is determined by credentials, permissions, and protocol support. The authenticated model catalog at /v1/models?capability=codex_responses is the authority for available Codex-compatible models. The Modal route supports native Responses forwarding, which allows Codex-style clients to use advertised models without changing the client workflow.

2. Account-Level Routing Policy and Scoped Router Keys: Routing policy lives in the Zerg account rather than inside each coding tool. Users can use provider API credentials with scoped Router keys, choose models, set limits, and configure routing behavior centrally. This architecture means a developer can switch models, enforce access rules, or update fallback behavior without editing configuration files in Codex, OpenCode, Pi, ZTC, or other clients. Scoped Router keys separate authentication from provider credentials, so API keys and personal subscriptions remain separate access paths. The router key authenticates routed requests, while provider credentials and Codex subscriptions remain distinct. This separation helps teams manage access without sharing primary provider keys across every tool.

3. Per-API-Key Daily Budgets Enforced Per Request: Zerg Router allows every API key to have its own daily budget, and that budget is enforced per request. This gives developers granular cost control across projects, teams, agents, or environments. A budget can prevent runaway usage from a single coding agent, test suite, or automation script. The enforcement happens at the routing layer before the request is forwarded, so the policy applies consistently across supported clients. The product also references account-wide caps and monthly usage caps for certain routes, including a $0.50 UTC-day account-wide trial cap and a 5M-token monthly quota for the eligible DeepSeek 4.1 Flash trial. Paid usage is metered, and bring-your-own provider keys are billed by the provider.

4. Explicit Model Fallback Chains for Reliability: Zerg Router supports explicit fallback chains for supported chat requests. If a primary route returns a 5xx error, a 429 rate-limit response, or a timeout, the router can try the next configured model. This improves reliability for coding agents that need continuous access to model inference. Instead of relying on a single provider or model, developers can define deterministic fallback behavior in the account-level routing policy. The fallback chain is explicit rather than hidden inside each tool, which makes routing behavior easier to audit and adjust. This feature is especially useful when using shared endpoints, trial capacity, or multiple providers with different rate limits and availability characteristics.

5. Codex Capacity Monitoring and Quota Freshness: Zerg Router provides visibility into connected Codex accounts by showing remaining weekly quotas, each quota’s reset date, and banked resets. Users can compare accounts by capacity and decide which account to use for a local Codex session. Router does not automatically switch a local Codex account; the user chooses the account. Quota data freshness is visible, and stale data is clearly marked. Account-wide Codex history does not invent missing token breakdowns. This helps developers avoid guessing about weekly limits, reset timing, or account capacity when running Codex-based coding agents.

6. Routed Usage Analytics and Transparent Token Reporting: Zerg Router lets users explore routed usage by model, date, and client. Where providers report them, users can see input tokens, cached input tokens, and output tokens. When token breakdowns are missing or unknown, the product does not fabricate them. This transparency is important for cost attribution, model comparison, budget planning, and debugging agent behavior. Usage analytics apply to routed requests, giving teams a centralized view of how models are being consumed across supported coding tools. The combination of per-key budgets, usage history, and quota monitoring creates a practical control plane for AI coding agent operations.

7. Flexible Funding, BYOK, and DeepSeek 4.1 Flash Access: Zerg Router supports paid usage on platform tokens or bring-your-own provider keys. Sign-up and Codex account monitoring are free. Eligible new personal accounts can try DeepSeek 4.1 Flash through Modal Shared Endpoints for 14 days after account creation, using a Router key, under a $0.50 UTC-day account-wide trial cap and a 5M-token monthly quota. Other Modal models are not included in that trial. After the trial, the paid DeepSeek route is metered at $0.30 per 1M input tokens, $0.03 per 1M cached input tokens, and $1.20 per 1M output tokens, up to the monthly usage cap. Beyond the trial, DeepSeek on platform tokens requires a $5/month plan, metered usage, and a card on file. Trial access also depends on shared platform capacity and Modal endpoint availability, so users should check their authenticated catalog before connecting a client.

Problems Solved

1. Pain Point: AI coding agents often require model configuration, API keys, budgets, and fallback logic to be repeated inside each tool. This creates fragmented routing policies, inconsistent cost controls, vendor lock-in, and operational risk when a provider returns 5xx errors, 429 rate limits, or timeouts. Developers may also lack visibility into Codex weekly quotas, reset dates, banked resets, and quota freshness across multiple accounts. Without centralized usage analytics, teams struggle to attribute model spend by API key, project, model, date, or client.
2. Target Audience: Zerg Router is built for software engineers, AI coding agent users, platform engineers, DevOps teams, indie developers, and engineering managers who use Codex, OpenCode, Pi, ZTC, or OpenAI-compatible clients. It is relevant for developers who want to run DeepSeek 4.1 Flash inside Codex without changing their workflow, teams that need per-API-key daily budgets, and organizations that bring their own provider keys. It also serves users managing multiple Codex accounts and those who need explicit fallback chains for production-grade agent reliability.
3. Use Cases: A developer can run DeepSeek 4.1 Flash inside Codex through one Router endpoint while keeping the Codex CLI workflow. A team can assign each project or agent its own API key with a daily budget enforced per request. A platform engineer can configure an explicit fallback chain that tries a secondary model on 5xx, 429, or timeout. A user with multiple Codex accounts can compare remaining weekly quotas, reset dates, and banked resets before starting a local session. An OpenCode, Pi, or ZTC user can connect through the same OpenAI-compatible provider. A cost-conscious builder can bring provider API keys or use paid metered platform usage. A team can review routed usage by model, date, and client to understand input, cached input, and output token consumption where reported.

Unique Advantages

1. Differentiation: Zerg Router differs from tool-specific model configuration and basic API proxies by placing routing policy in the account rather than inside each coding tool. It combines an OpenAI-compatible endpoint, scoped Router keys, per-API-key daily budgets, explicit model fallback chains, Codex quota monitoring, and routed usage analytics in one layer. It supports both paid platform usage and bring-your-own provider keys, so users are not forced into a single billing path. It also preserves existing clients such as Codex, OpenCode, Pi, ZTC, and other OpenAI-compatible apps, reducing migration effort. Transparent quota freshness and non-fabricated token reporting further separate it from opaque routing tools.
2. Key Innovation: The key innovation is account-level routing policy with per-request budget enforcement and explicit fallback behavior. A Router key authenticates routed requests, while provider credentials and Codex subscriptions remain separate access paths. The router can forward supported requests to models such as modal/deepseek-v4.1-flash and use native Responses forwarding for Codex-compatible workflows. By exposing quota reset dates, banked resets, and data freshness across connected Codex accounts, Zerg Router turns model access, cost control, and capacity planning into a centralized developer control plane. The authenticated catalog remains the authority for available models, ensuring users connect only to routes their credentials and permissions support.

Frequently Asked Questions (FAQ)

1. What is Zerg Router? Zerg Router is an OpenAI-compatible API router and developer application for coding agents. It provides one endpoint at https://zergrouter.com/v1 for tools such as Codex, OpenCode, Pi, ZTC, and other OpenAI-compatible apps. It centralizes model choice, provider credentials, scoped Router keys, per-API-key daily budgets, explicit fallback chains, Codex quota monitoring, and routed usage analytics.
2. Can I use DeepSeek 4.1 Flash with Codex? Yes. Use a model advertised by /v1/models?capability=codex_responses. The Modal route supports native Responses forwarding, and the authenticated catalog is the authority for available models. Eligible new personal accounts can try DeepSeek 4.1 Flash through Modal Shared Endpoints for 14 days after account creation, under a $0.50 UTC-day account-wide trial cap and a 5M-token monthly quota. Other Modal models are not included.
3. Does a Zerg Router API key include a model subscription? No. A Router API key authenticates routed requests. Provider credentials and Codex subscriptions remain separate access paths. New personal accounts have only the bounded Modal DeepSeek trial described by the platform. Beyond the trial, DeepSeek on platform tokens requires a $5/month plan, metered usage, and a card on file. Bring-your-own provider keys are billed by the provider.
4. How are Zerg Router requests funded? Sign-up and Codex account monitoring are free. Eligible new personal accounts receive a bounded 14-day DeepSeek trial on platform tokens with a $0.50 UTC-day account-wide cap and a 5M-token monthly quota. The paid DeepSeek route is metered at $0.30 per 1M input tokens, $0.03 per 1M cached input tokens, and $1.20 per 1M output tokens, up to the monthly usage cap. Own provider keys are billed by the provider.
5. Will Zerg Router automatically switch my local Codex account? No. You choose the account for a local Codex session. Zerg Router shows remaining weekly quotas, reset dates, banked resets, and quota data freshness across connected Codex accounts so you can decide which account to use. It does not automatically switch the local Codex account.
6. How do I get access to Zerg Router? Continue to ZergAI to check current account access at https://zergai.com. Existing users can sign in. Access and any billing requirements are confirmed in ZergAI rather than inferred from the product page. Available models depend on credentials, permissions, protocol support, shared platform capacity, and the authenticated catalog.