Features

Everything that adds up.

LLM N8 is a local-first console for the AI accounts you already own: usage sync, dashboards, budget guardrails, alerts, model comparison and a complete record you can export at any time.

LLM N8 home tab with totals, budgets and anomalies

Overview

One home for every account.

Token totals, cost, request counts and model counts sit above the fold, with a provider filter for narrowing the picture. Below them come budgets, anomaly alerts and the token mix.

Aggregate or per-account totals
Configurable timeframe, from today to a year
Budgets and alerts surfaced where you land first

Dashboards

From headline number to raw rows.

Every dashboard is editable. Change a widget's metric, scope it to accounts or models, rename it, reorder it, or drill into the usage rows behind a chart.

Accounts, models, tags and date ranges as filters
Saved segments for the views you reuse
Annotations for launches, migrations and price changes
CSV and PNG sharing from any dashboard

Dashboard details with token charts in LLM N8

Models

Pick a model with the price in view.

Synced catalogs carry context size and per-million token pricing. Compare candidates side by side, then run the same prompt across them and compare the real answers, latency and cost.

Searchable catalog per account
Pricing and context-window comparison
Batch runs with per-model results

Six dashboards to start from

Create a dashboard in seconds from a template, then make it yours. Every template is a normal editable dashboard — nothing is locked.

Overview

Totals, mix and the models driving them.

Trend

Usage and spend over time, with a period comparison.

Budgets

Spend against limits and month-end projection.

Key spend

Which account and key are responsible for the bill.

Latency & reliability

p50/p95/p99 latency, error rate and fallbacks.

Live

A short-window view that refreshes as usage arrives.

Eleven widget types

Mix and match the widget that fits the question instead of forcing every metric into the same chart.

KPI tile — one headline number, with a period comparison.
Goal gauge — progress against a spend cap or goal.
Time series — tokens, cost, requests, latency or errors over time.
Bar chart — breakdown by model, provider, tag or fallback target.
Donut chart — token mix or share by model.
Usage table — the underlying rows, sortable and readable.
Period comparison — this window against the one before it.
Trend line — cost, requests or model count with a fitted direction.
Request heatmap — when the requests actually happen.
Stacked bars — composition of tokens or cost by model.
Tokens & requests — the two series that explain each other.

Budgets that warn you early

Set a monthly or weekly limit per account and LLM N8 tracks spend against it. Alerts are evaluated locally the moment new usage lands.

Threshold alerts before the limit is reached
Projected-overrun warnings based on the current pace
Low-credit alerts for providers that publish a balance
Anomaly detection using median absolute deviation, so one unusual day stands out from normal variance

Accounts without a usage API are called out explicitly and excluded from budget maths rather than silently showing a false zero.

Your data, in your hands

Accounts, models, usage history, budgets and dashboards live in a local SQLite database. API keys live in the operating system keychain. There is no LLM N8 server to sync with.

Export everything as JSON, or usage rows as CSV
Optional passphrase-encrypted backup envelopes
Import an export to restore or move to a new device
Remove an account to delete its keys and cascade its history
Reset all data to wipe the database, keys and preferences

23 languages, local formats

The interface ships in 23 languages, from English and Español to 日本語 and Tiếng Việt, with locale-aware numbers and dates. The language picker lives in Settings and can be changed at any time.

No account, no telemetry

LLM N8 has no sign-up and no analytics. The only network requests it makes are to the providers you connect: your own API key, your own usage endpoint. Read the privacy policy for the full detail, including what is stored and how to delete it.

Launching soon

Watch the bill, not the meter.

Coming soon to the App Store and Google Play.

Browse screenshots Check provider support