Now in early access — free plan available

Know exactly what your
AI agents cost.

Real-time cost tracking, forecasting, and anomaly detection for LLM-powered agents. Drop in the Python SDK or point your agent at the bridge, and see spend by agent, model, and day.

Start tracking for free View pricing

No credit card required for the free plan.

Built for agent teams

Everything you need to keep LLM spend predictable, visible, and under control.

Cost by agent & model

See per-agent spend and model-level breakdowns. No more hunting through provider invoices.

Forecasting & anomalies

Project monthly spend and catch cost spikes before they surprise finance.

One-line SDK & bridge

Patch OpenAI or Anthropic clients in one line, or point closed binaries like Claude Code at the bridge. Events stream to your dashboard in real time.

Simple, predictable pricing

Start free. Upgrade when you're running agents in production.

Free

For solo builders and early validation.

$0 /month

Up to 1 agent
Real-time cost dashboard
7-day cost history
Python SDK & bridge

Get started free

Pro

For teams running agents in production.

$29 /month

Up to 5 agents
30-day cost history
Cost forecasting & anomaly alerts
Priority support

Upgrade to Pro

Two ways to instrument everything

Patch the Python SDK in next to your existing OpenAI, Anthropic, or Ollama client — or point a closed binary like Claude Code at the bridge. No rewrites.

Python SDK

Monkey-patch the official clients in one call. Supports OpenAI, Anthropic, and Ollama.

from agentmonitor import patch_openai
patch_openai()  # existing calls are now tracked

Download & docs on GitHub

↔

Bridge proxy

For closed binaries you can't patch — Claude Code, Cursor, Aider, Continue. Point the agent's base URL at it; it forwards and records usage. Zero deps, Node 18+.

agent-monitor-bridge init
export ANTHROPIC_BASE_URL=http://localhost:8787
claude  # calls are now tracked