Install
Run one command. The Systray app installs on your laptop, detects your coding tools, and points them at the router. Add your API keys (BYOK) or use Monk Cloud's managed models.
$ npx @themonk-ai/install
Monk Router is a plug-and-play proxy that routes every request to the cheapest model that meets your quality bar. Drop-in for Anthropic, OpenAI, and Gemini wire formats. No code changes.
Pick a quality preset and your monthly LLM spend. Monk Router routes each request to the cheapest model that clears your quality floor — typically 40-70% less cost, no quality regression.
Estimates use a 40-70% savings range depending on quality preset and traffic mix. Actual savings vary by workload. No savings = $0. You only pay when Monk saves you money.
A router, not a gateway. Quality-aware routing, not just provider failover. Built for teams that refuse to overpay for inference.
Every request is scored and routed to the cheapest model that clears your quality floor. If no cheaper model meets the bar, it stays on the requested model — zero quality regression, zero risk.
Drop-in proxy speaks Anthropic, OpenAI, and Gemini wire formats. Point your base URL at Monk Router and start saving.
Every routed response is scored against the original. If quality would regress below your floor, Monk Router keeps the requested model. No degradation, ever.
Customer-hosted data plane. Prompts and API keys never leave your infrastructure.
Bring your own keys, or let Monk Cloud resell OpenRouter's 100+ models at cost.
Dial savings up or down per org, team, or user. Spend caps, deny lists, tier ceilings.
VPC deployment (ECS Fargate), SSO, admin controls, audit log, RBAC. Built for compliance-sensitive environments.
Three steps from install to savings. No code changes, no infra to manage.
Run one command. The Systray app installs on your laptop, detects your coding tools, and points them at the router. Add your API keys (BYOK) or use Monk Cloud's managed models.
Each request is embedded, matched to a cluster, and scored against every candidate model. The router picks the cheapest model whose quality score clears your floor — in under 40ms.
Pay only for savings. Monk takes a percentage of what you save — if there are no savings, you pay $0. Track everything in the dashboard: per-request savings, model mix, quality scores, routing health.
Whether you're an individual dev or an enterprise team, there's a Monk Router deployment for you.
Runs on your laptop. Bring your own API keys. $0 if no savings.
Managed SaaS. Monk resells OpenRouter (100+ models). $500/mo min.
Runs in your VPC. BYOK, SSO, admin controls. Negotiable.
Gateways route by price. Monk Router routes by quality-per-dollar — and keeps your data in your infrastructure.
| Capability | Monk Router | OpenRouter | Portkey | Not Diamond |
|---|---|---|---|---|
| Quality-aware routing (per-prompt scoring) | ✓ | — | — | ✓ |
| Quality floor (no regression guarantee) | ✓ | — | — | — |
| Customer-hosted data plane (VPC) | ✓ | — | partial | — |
| BYOK (your keys never leave) | ✓ | — | ✓ | ✓ |
| Cost-vs-value regulators (per-org/user dials) | ✓ | — | — | — |
| Drop-in proxy (Anthropic / OpenAI / Gemini wire formats) | ✓ | ✓ | ✓ | ✓ |
| Spend caps + deny lists + tier ceilings | ✓ | partial | ✓ | — |
| Percentage-of-savings pricing | ✓ | — | — | — |
| SSO / SAML / audit log | ✓ | — | ✓ | — |
You only pay when Monk Router saves you money. No upfront costs, no per-seat licenses. If there are no savings, you pay nothing.
| Product | Monk's share | Minimum | Usage pass-through |
|---|---|---|---|
| Monk Edge | 25% of savings | $0 (no minimum) | BYOK — your keys, your costs |
| Monk Cloud | 30% of savings | $500/mo minimum | OpenRouter pass-through at cost |
| Monk Enterprise | 25% of savings (negotiable) | Custom | BYOK — your keys, your costs |
Savings = (requested cost) − (actual cost). Monk's share is calculated per-request and aggregated monthly. No savings = $0.
Monk Router speaks to every stakeholder in the LLM spend conversation.
"I want cost control without a developer revolt."
Drop-in proxy, no code changes, per-user presets. Developers don't feel it; the bill goes down.
"I want forecastable spend and demonstrable savings."
Spend caps, per-org regulators, monthly savings reports. Percentage-of-savings means aligned incentives.
"No prompt exfiltration. Full auditability."
Customer-hosted data plane, BYOK, SSO, audit log. Prompts and keys never leave your network.
"Don't add latency. Don't drop quality."
Sub-40ms routing overhead. Quality floor guarantees no regression. Cache hits are instant and free.
"Operable, HA, clear on-call story."
Containerized data plane, health endpoints, failover, provider 503 handling. Runs in ECS Fargate or on a laptop.
Deploy in your VPC with full control over keys, access, and audit.
The data plane runs as a container in your VPC (ECS Fargate or EC2). Your API keys never leave your network. SSO, admin controls, and audit logs for compliance-sensitive environments.
Contact salesInstall Monk Router, add your keys, and start cutting LLM costs today. No code changes required. No savings, no charge.
Trusted by teams cutting LLM spend
From individual developers to enterprise platform teams.