Know what every
n8n workflow costs.
TokenSense tracks what every n8n AI call costs. Change one URL in your credentials, and every request gets logged with its exact cost, token count, and the workflow that made it. Works with OpenAI, Anthropic, Google Gemini, xAI, and Mistral.
No credit card · 10k requests/mo free · 5-min setup
The problem
Your n8n AI costs are a black box.
The meter is always running.
A workflow looped on GPT-4. No caps, no alerts, no kill switch. Just a bill you found on Monday.
OpenAI shows one number. Anthropic another. No breakdown by workflow or project.
One key shared across every n8n workflow. One rate limit takes everything down.
Your CFO asks what the Lead Qualifier costs to run. You don't know. Nobody does.
Before & After
One URL change. Full visibility.
- No per-workflow costs
- No budget caps
- No usage logs
- One shared API key
- Cost per workflow, per model
- Budget caps that block spend
- Every request logged
- Per-project API keys
Get started
Two ways to connect.
Choose the setup that fits your workflow.
Swap your base URL
Best for: Quick start, any existing workflow
- No install required
- Works with any AI node
- 2-minute setup
- Change one URL and go
Install the community node
Best for: Full-featured cost tracking
- Native model dropdowns in n8n
- Per-step cost metadata on every output
- Built-in budget controls
- AI Agent sub-node for LangChain workflows
Community node
Native AI controls, right inside n8n.
The TokenSense community node gives you model selection, cost metadata, and budget controls — all without leaving n8n.
Native model dropdowns — no memorizing model IDs
Setup
Running in five minutes.
Copy the endpoint below and paste it into your n8n credentials.
https://api.tokensense.ioIn n8n: Credentials → OpenAI → Base URL
Same endpoint for every workspace — just paste and go.
How it works
Three steps, zero friction
Sign up & grab your endpoint
Create a free TokenSense workspace. You get your endpoint URL and API key instantly — no approval, no waitlist.
Add your provider keys
Paste your OpenAI, Anthropic, or Gemini keys into TokenSense Settings. Encrypted with AES-256-GCM — only used to forward your requests.
Swap the URL in n8n
In your n8n credential settings, change the base URL to your TokenSense endpoint. That's it — your workflows run exactly the same.
What you get
Full visibility, zero code
Per-workflow cost tracking
See exactly what each workflow spends on AI calls. No tagging, no custom headers — automatic attribution.
Budget caps that enforce
Set monthly or daily limits per workspace or project. When the cap is hit, requests stop — not just an alert.
Multi-provider support
Works with OpenAI, Anthropic, Gemini, Mistral, and xAI. Use any provider through TokenSense without changing your workflows.
Real-time usage logs
Every AI call logged with tokens, cost, model, latency, and status. Filter by workflow, project, model, or time range.
Under 50ms overhead
Parallel enforcement checks and async logging. Your workflows stay fast — TokenSense adds minimal overhead.
Smart alerts
Budget thresholds, error spikes, latency degradation, and spend anomalies. Get notified before the bill surprises you.
Your dashboard
See exactly where the money goes.
Every workflow, every model, every dollar — in one place.
Compatibility
Works with the nodes you already use
Any n8n node that lets you set a custom base URL is compatible. Here are the most common ones.
Learn more
Guides for n8n teams
FAQ
Common questions
TokenSense sits between your n8n workflows and your AI providers. You change the base URL in your n8n AI credentials to point at TokenSense instead of the provider directly. TokenSense forwards requests, logs everything, enforces budgets, and sends responses back. Your workflows run exactly the same — they just gain full cost visibility.
TokenSense works with n8n's built-in OpenAI, Anthropic (Claude), and Google Gemini nodes, plus the AI Agent Chat Model sub-nodes. It also works with HTTP Request nodes making AI API calls. Any node that lets you set a custom base URL is compatible.
Minimal — typically under 50ms per request. Enforcement checks (auth, quota, budget) run in parallel, and logging happens asynchronously after the response is returned to your workflow.
Yes. You add your own OpenAI, Anthropic, or Gemini keys to TokenSense. They're encrypted with AES-256-GCM and only used to forward your requests. You can also set up per-project keys if you manage multiple clients.
No. The base setup only requires changing a URL and API key in your existing n8n credentials. We also offer an optional community node (n8n-nodes-tokensense) with extra features like dynamic model dropdowns and metadata injection, but it's not required.
Yes. The Starter plan is free and includes 10,000 AI requests per month, full cost tracking, and budget caps. No credit card required.
See what your n8n workflows actually cost.
Free to start. 10,000 requests/month on the Starter plan. No credit card required.
No credit card required · 5-minute setup