The Analytics Layer of AI

Carbon turns every LLM and tool call into a queryable ledger — sliced by agent, model, user, and thread — so engineering and product decisions stop running on vibes.

Carbon App

Find answers
to every question

You shouldn’t have to play detective with your own AI. The invoice arrives as one number, the logs scroll past a million rows deep, and the questions that decide your roadmap go unanswered. Carbon keeps every call in a queryable ledger, so the answer is a click — not an investigation.

What changed after Tuesday's deploy?
Why did our AI bill double this month?
Which agent burned $300 last Tuesday?
Which tool calls keep failing?
Which model handles long context cheapest?
Which threads run the longest?
Who are our heaviest users, and what do they cost us?
Which model is actually cheapest for this task?
Are retries double-counting our usage?
Where do agents stall mid-run?
Which provider failed most this week?
What did that support thread cost?
Is the new prompt actually cheaper?
Which feature is driving token spend?
What's our cost per active user?

One ledger for every provider

  • OpenAIOpenAI
  • AnthropicAnthropic
  • GoogleGoogle
  • VercelVercel
  • MistralMistral
  • OpenRouterOpenRouter
  • DeepSeekDeepSeek
  • GrokXAI
  • KimiKimi
  • AlibabaAlibaba
  • MinimaxMiniMax
  • Xiaomi

Dive deep into clusters

Events are grouped into clusters — scoped to your space and time range, ready to drill, whether the question is an engineering one or a product one.

  • Agents

    Which agents run, what they cost, what they touch

  • Cost

    Spend over time, and exactly what is driving it

  • Models

    Calls and spend by model, across every provider

  • Users

    The humans behind the threads, ranked

  • Tools

    Call volume, failure rate, and duration for every tool

Integrate in
two lines.

Use our SDK to capture events from your existing AI providers. All it takes is two lines of code and your events will start appearing in your dashboard.

import OpenAI from "openai";
import { Carbon } from "@carbon-js/sdk";
import { wrapOpenAISdk } from "@carbon-js/sdk/ai";

// Wrap your client with Carbon SDK.
const carbon = new Carbon();
const openai = wrapOpenAISdk(new OpenAI(), carbon);

// Use exactly as before — every call is captured.
const completion = await openai.responses.create({
  model: "gpt-5.5",
  input: "Hello, Carbon!",
});

Start free.
Scale when you're ready.

Free forver. No credit card. No time limit.

Carbon Free

$0/ month

For personal projects

Features

1,000 events per month

Additional events not supported

LLM and tool call tracking

All providers and SDKs supported

All clusters included

Carbon Plus

$28/ month

For production projects

Features

100,000 events per month included

$0.00028 per additional event

LLM and tool call tracking

All providers and SDKs supported

All clusters included

Questions?

Everything here describes how Carbon works today — not a roadmap.

What counts as an event?

One captured LLM call or one tool call.

Which providers and SDKs are supported?

Our SDK supports all major providers and SDKs, including OpenAI, Anthropic, Vercel, and more.

How much code does it take to try?

It takes two lines of code to get started with your existing AI providers.

What happens if my included usage is reached?

If you are on the Free plan, you will not be able to capture events until the next period starts. You can upgrade to a Plus plan to continue capturing events. If you are on the Plus plan, events will be captured as usual, and you will be charged for the overage.

Do you store prompts and outputs?

Yes — the full event, including inputs and outputs, is stored in our database. We are working on a way to configure this.

Can I include custom fields in the events?

Yes, you can include custom fields in the events. This is useful for tracking additional metadata about the events that you want to analyze.

Can I take my data out of Carbon?

Yes, you will be able to export the events very soon. We are working on a way to implement this.

~~~ Let's begin ~~~

Two lines of code, a thousand free events a month — and every decision after that comes with evidence.