Question 1

What is a Token Plan?

Accepted Answer

A Token Plan is your account's single subscription on Router9 — one flat monthly fee with its own credit quota, rate limits, and audit trail. You can issue multiple API keys under it (one per harness or environment). A harness can be an autonomous agent or a coding CLI.

Question 2

How is Router9 different from OpenRouter?

Accepted Answer

OpenRouter bills per token from a prepaid credit balance, so costs are unpredictable. Router9 is a flat-rate subscription gateway for any AI harness: a fixed monthly price with clear rate limits instead of per-token metering. It fronts multiple providers for the best performance-per-dollar, rolls unused credits into next month (60% on Pro, 80% on Max, up to 2× your quota), and bundles voice, vision, web search, image generation, and audit logging that you'd otherwise stitch together from separate providers.

Question 3

Does Router9 work with Hermes, OpenClaw, and Nanobot?

Accepted Answer

Yes — you can run Hermes, OpenClaw, Nanobot, Goose, Letta, and other agent harnesses on a flat monthly plan without per-token billing. Router9 works with any AI harness that speaks the OpenAI chat completions format. Point the harness's base URL at https://api.router9.com/v1 and use your sk-r9k- key; a native Anthropic Messages endpoint is also available.

Question 4

Do credits roll over to the next month?

Accepted Answer

Yes, on paid plans. Each plan grants a monthly credit quota, and you keep a share of whatever you don't use into the next month — 60% on Pro and 80% on Max (the Free tier does not roll over). Rolled-over credits accumulate up to a hard cap of 2× your monthly quota — so quiet months build a buffer for bursty ones, and your effective allowance is always whatever you've banked plus this month's grant.

Question 5

Do you charge per token used?

Accepted Answer

No. Router9 uses a flat monthly subscription per Token Plan. Each plan includes a generous monthly credit quota that meters usage by per-model cost — so you never see a per-token invoice — and on paid plans unused credits roll over (60% on Pro, 80% on Max, up to 2× quota). If you exceed your effective allowance you can top up pay-as-you-go, but for most harnesses the fixed fee is the whole bill: a predictable line item, not a variable.

Question 6

Which AI models can I access?

Accepted Answer

Every major provider through a single endpoint — OpenAI (GPT), Anthropic (Claude), Google (Gemini), DeepSeek, and hundreds more. Router9 fronts multiple providers and routes each request for the best value. Set the model to "auto" and Router9 picks the best model for your plan — on the Free tier it draws only from free-eligible models, so "auto" always lands on something you can use. Pro and Max unlock every model. (Explicitly naming a premium model on the Free tier still returns 403 model_requires_upgrade.)

Question 7

Can I give each harness its own budget and detect leaked keys?

Accepted Answer

Yes — Router9 treats each API key as one harness identity with its own controls (Pro and Max plans). Give a key a monthly credit budget with a soft alert (email at 80% by default) or a hard cap that returns 429 once exhausted. Restrict it to specific models with a model allowlist, or set a per-key rate limit. Bind a key to the harness you expect it to run: if its traffic later classifies as a different tool, Router9 emails you — a strong signal the key leaked or was shared. Drift detection is alert-only and never blocks traffic; if a key is compromised, Refresh the key to rotate the secret in place while keeping its name, budget, policies, and full history.

Question 8

What can my harness do beyond chat completions?

Accepted Answer

Router9 ships built-in Skills: speech-to-text, text-to-speech, image understanding, image generation, and real-time web search — with video generation and more on the roadmap. Install the Router9 Skill package (or MCP server) and your harness gains ears, a voice, and eyes — no extra SDKs or providers. Every tier includes every Skill; quotas scale with your plan.

Question 9

What happens if my Token Plan hits the rate limit?

Accepted Answer

The API returns a 429 Too Many Requests response with a Retry-After header showing exactly when you can retry. (Total monthly volume is governed separately by your credit quota plus any rolled-over credits.) Your harness can retry after that interval, or you can upgrade the plan at any time for higher limits — new limits take effect immediately.

The Unified Token Plan for AI Harnesses

Build Your Own Agent Team

Every Model, One Endpoint

Context-based Auto-Routing

Skills, Built In

Drop-In Compatible

Secure Every Harness

Observability, By Harness

Flat Monthly Pricing

Free

Pro

Max

Frequently Asked Questions

Give every AI harness a key you control