Introducing Manifest

Team Manifest Team Manifest 3 min read
Manifest dashboard showing cost savings across LLM providers

The Problem with LLM Costs

How many users were surprised when they saw their AI consumption bill after playing a bit with OpenClaw?

Personal AI assistants like OpenClaw use provider API keys where you pay per usage. But estimating the cost of each request is very hard.

What Manifest Does

Manifest is a cost-saving platform for OpenClaw users. Manifest sits between your agent and the LLM providers. It intercepts every request and routes it to the most cost-effective model to reduce your costs by up to 70%. No code changes required — just install the plugin and go.

Here’s what that looks like in practice:

  • Simple queries (classification, extraction, formatting) get routed to smaller, cheaper models
  • Complex queries (reasoning, code generation, analysis) go to frontier models
  • Everything is tracked — you see exactly what each request costs, in real time

How It Works

Manifest uses a lightweight routing engine that analyzes the complexity of each request. It considers:

  1. The type of task (code, text, classification, etc.)
  2. Required quality level based on the prompt
  3. Current model pricing across providers
  4. Your configured preferences and constraints

The router makes a decision in under 2ms and sends it to the right model.

Your data stays yours

We believe that LLM Routing is going to play a big role in the upcoming years as it’s a low-hanging fruit to reduce AI consumption. We also believe that users should own this segment instead of trusting opaque and proprietary proxies that have access to the content of your messages and can have biases when choosing the provider.

Manifest cares about your data. The complexity scoring is 100% done locally. On the local version, the routing is too. On the cloud version, we use a Blind Proxy that redirects the query to the right provider without seeing it by architecture.

The code is open source and available on GitHub.

Real-Time Cost Tracking

Beyond routing, Manifest gives you full visibility into your AI spending. The dashboard shows:

  • Per-message costs — see exactly what each interaction costs
  • Per-model breakdown — compare spending across providers
  • Usage trends — track costs over time and spot anomalies
  • Budget alerts — set limits before costs get out of hand

All of this runs locally on your machine. Your data never leaves your environment.

Getting Started

Getting started takes about 30 seconds. Log in to the Manifest Cloud Platform, get your agent API key, and install the plugin:

openclaw plugins install manifest

Configure it in your OpenClaw settings, and you’re done. Manifest handles the rest.

What’s Next

We’re just getting started. On the roadmap:

  • Unified API key for accessing all models with one key
  • Custom routing rules for fine-grained control
  • Provider failover for automatic retry on errors

We’d love your feedback. Join us on Discord or check out the GitHub repo.