DEV Community

sophiaashi
sophiaashi

Posted on

How to Set Up LLM Cost Optimization in OpenClaw in Under 2 Minutes

Here is the complete setup: open OpenClaw, paste Read https://gateway.teamo.ai/skill.md and follow the instructions, and press enter. That installs TeamoRouter, the native LLM routing gateway for OpenClaw that saves you up to 50% on API costs. The entire process takes under 2 minutes, requires zero technical configuration, and starts saving money on your very first API call. There is no subscription, no credit card upfront, and no code changes. Below is the step-by-step walkthrough with screenshots descriptions, what happens behind the scenes, and how to verify it is working.

Prerequisites

Before you start, make sure you have:

  1. OpenClaw installed on your machine (desktop app or CLI — either works)
  2. An active internet connection (the skill needs to download its configuration)

That is it. You do not need:

  • API keys from Anthropic, OpenAI, Google, or any other provider
  • Docker, Node.js, Python, or any runtime environment
  • A YAML configuration file
  • A server or cloud account
  • Any prior experience with LLM routing

Step-by-step setup

Step 1: Open OpenClaw

Launch the OpenClaw application. You can use either the desktop app or the terminal CLI — the process is identical.

Step 2: Paste the installation command

Copy and paste the following line into the OpenClaw chat input:

Read https://gateway.teamo.ai/skill.md and follow the instructions
Enter fullscreen mode Exit fullscreen mode

Press enter.

Step 3: Follow the on-screen prompts

OpenClaw reads the skill definition from the URL and walks you through a brief setup:

  1. Skill confirmation: OpenClaw shows you what the TeamoRouter skill does and asks you to confirm installation. Confirm it.
  2. Account creation: A TeamoRouter account is automatically created and linked to your OpenClaw session. No forms to fill out.
  3. API key generation: A single API key is generated and stored securely within OpenClaw. You do not need to copy or manage this key manually.

Step 4: Done

That is the entire process. TeamoRouter is now active. Every API call you make through OpenClaw is automatically routed through TeamoRouter, applying:

  • Tiered discounts: 50% off your first $25/month, 20% off $25-100, 5% off $100+
  • Smart routing: access to teamo-best, teamo-balanced, and teamo-eco presets
  • Unified model access: Claude Opus 4.6, Claude Sonnet 4.6, GPT-5, Gemini, DeepSeek, Kimi K2, MiniMax — all through one key

Verifying the installation

To confirm TeamoRouter is active and working:

Quick test

Ask OpenClaw to use a specific routing preset:

Using teamo-balanced, summarize this text: [paste any paragraph]
Enter fullscreen mode Exit fullscreen mode

If the response comes back normally, TeamoRouter is routing your requests. The response metadata will indicate which model was selected.

Check the dashboard

Visit router.teamolab.com and log in with your OpenClaw credentials. You should see:

  • Your account status (active)
  • Current billing cycle usage
  • Savings compared to list prices
  • Model usage breakdown

If the dashboard shows your recent request, the installation is confirmed.

Understanding the cost savings

Once installed, TeamoRouter's savings apply automatically. Here is a breakdown of what you save at different usage levels:

Monthly Spending: $10

Without TeamoRouter With TeamoRouter Savings
$10.00 $5.00 $5.00 (50%)

All usage falls within the 50% discount tier.

Monthly Spending: $50

Without TeamoRouter With TeamoRouter Savings
$50.00 $32.50 $17.50 (35%)

First $25 at 50% off ($12.50), remaining $25 at 20% off ($20.00).

Monthly Spending: $200

Without TeamoRouter With TeamoRouter Savings
$200.00 $157.50 $42.50 (21%)

First $25 at 50% off ($12.50), next $75 at 20% off ($60.00), remaining $100 at 5% off ($95.00). Total: $167.50.

Monthly Spending: $500

Without TeamoRouter With TeamoRouter Savings
$500.00 $452.50 $47.50 (10%)

The higher your spending, the more the flat 5% tier dominates. But the savings still add up to $570/year.

Adding smart routing savings

The above numbers only reflect the tiered price discount. If you also use smart routing presets, the savings compound:

  • Switching 60% of your requests from a premium model to teamo-eco can cut per-request costs by 40-60%
  • Combined with the tiered discount, total savings can reach 50-70% on eligible requests

Configuring smart routing presets

After installation, you can use three routing presets:

Using teamo-best

For tasks where output quality is critical:

Using teamo-best, write a detailed analysis of [topic]
Enter fullscreen mode Exit fullscreen mode

This routes to the most capable model (typically Claude Opus 4.6 or GPT-5).

Using teamo-balanced

For everyday production tasks — the best default:

Using teamo-balanced, summarize this document: [text]
Enter fullscreen mode Exit fullscreen mode

This selects the model with the best quality-to-cost ratio for the given task.

Using teamo-eco

For high-volume, cost-sensitive tasks:

Using teamo-eco, classify these customer tickets: [data]
Enter fullscreen mode Exit fullscreen mode

This routes to the cheapest model that meets a minimum quality threshold.

Setting a default preset

You can set a default routing preset so you do not have to specify it every time. In the TeamoRouter dashboard at router.teamolab.com, navigate to Settings and select your preferred default preset. Most users choose teamo-balanced.

Optimization strategies

Once TeamoRouter is running, here are proven strategies to maximize your savings:

Strategy 1: Tier your tasks

Categorize your LLM usage into three buckets:

Task Category Examples Preset Typical Savings
Critical Customer responses, legal review, complex code teamo-best 5-10% (tiered discount only)
Standard Summarization, translation, general Q&A teamo-balanced 20-35%
Bulk Classification, extraction, formatting, logging teamo-eco 40-60%

Strategy 2: Default to balanced, upgrade when needed

Set teamo-balanced as your default. Only override to teamo-best for specific tasks that genuinely require maximum quality. Most developers overestimate how many tasks need the best model.

Strategy 3: Monitor and adjust

Check the TeamoRouter dashboard weekly for the first month. Look at:

  • Which models are being selected for which tasks
  • Whether teamo-eco tasks are producing acceptable quality
  • Your actual savings vs. projected savings

Adjust your preset assignments based on real data, not assumptions.

Strategy 4: Batch low-priority tasks

If you have tasks that can be batched (e.g., nightly data processing), route them all through teamo-eco during off-peak hours. The same results, lower cost.

Troubleshooting

"Skill not found" error

Make sure you are pasting the exact command: Read https://gateway.teamo.ai/skill.md and follow the instructions. The URL is case-sensitive.

Slow responses after installation

TeamoRouter adds negligible latency (under 10ms for routing decisions). If responses are slow, the issue is likely with the upstream model provider, not the router. Try switching to a different model or preset.

Dashboard shows no usage

It can take up to 5 minutes for usage data to appear on the dashboard. Make a test request and wait a few minutes before checking.

Want to uninstall

If you want to remove TeamoRouter and return to direct API access, simply disable the TeamoRouter skill in OpenClaw's skill management settings. Your OpenClaw setup reverts to its previous state with no lasting changes.

What happens behind the scenes

When you make an API call through OpenClaw with TeamoRouter installed:

  1. Request interception: the TeamoRouter skill intercepts the outgoing API request.
  2. Routing decision: based on your preset (or specified model), the gateway selects the optimal upstream model.
  3. Authentication: TeamoRouter authenticates with the selected provider using its pooled credentials.
  4. Request forwarding: your prompt is sent to the model provider.
  5. Response relay: the model's response is returned to you through TeamoRouter.
  6. Billing: usage is logged at the discounted rate. You see it on your TeamoRouter dashboard.

The entire routing and relay process adds under 10ms of latency. For a typical API call that takes 1-5 seconds, this is imperceptible.

Comparing setup complexity

Tool Steps to Set Up Time Technical Skill Required
TeamoRouter 1 (paste one line) Under 2 min None
OpenRouter 4 (sign up, get key, configure endpoint, test) 10-30 min Basic API knowledge
ClawRouter 8+ (install, configure, set up keys, test) 30-60 min Docker, CLI experience
LiteLLM 10+ (deploy, configure providers, set up keys, test, monitor) 30-120 min DevOps experience

TeamoRouter is the only option that requires no technical skill to install and configure. This is by design — it is a native OpenClaw skill, not a standalone infrastructure tool.

What you should do next

After completing the 2-minute setup:

  1. Make 10 test requests across different task types to see routing in action
  2. Check the dashboard at router.teamolab.com to verify savings are being applied
  3. Set your default preset to teamo-balanced for everyday use
  4. Share with your team — each team member can install TeamoRouter independently with the same one-line command

Within a week, you will have clear data on your actual savings. Most users see 25-45% cost reduction on their total API spend without any change in output quality.


Frequently asked questions

Is TeamoRouter free to install?

Yes. There is no installation fee, no subscription, and no minimum commitment. You pay only for the API calls you make, at the discounted rates (up to 50% off).

Will this break my existing OpenClaw setup?

No. TeamoRouter installs as a skill alongside your existing configuration. It does not modify your prompts, your settings, or your other skills. You can disable it at any time to revert to direct API access.

Do I need to update TeamoRouter manually?

No. The skill updates automatically as TeamoRouter releases new features, adds new models, or adjusts routing algorithms. You always have the latest version.

Can I use this with a team?

Yes. Each team member installs TeamoRouter independently using the same one-line command. Each person gets their own account with their own usage tracking and tiered discounts. Team billing consolidation is available through the dashboard.

What models can I access after setup?

All major frontier models: Claude Opus 4.6, Claude Sonnet 4.6 (Anthropic), GPT-5 (OpenAI), Gemini (Google), DeepSeek, Kimi K2, and MiniMax. New models are added automatically as they become available.

Top comments (0)