Marginal
Marginal is a cost-observability service for LLM usage that tracks and attributes the cost of every model call to fields you define (customer, feature, model) and computes pricing server-side so you never do token math in your app.
Marginal is business intelligence software teams evaluate for business intelligence. Use this page to review pricing, integration signals, and the best alternatives before you commit.
Profile facts come from the vendor source. AiMatch labels unknown pricing or API details instead of estimating them.
Review official source →Used in These Packs
Quick Overview
Best for: Business Intelligence
What it does
Business Intelligence software for decision-makers comparing workflow fit and alternatives.
Best fit
Business Intelligence
Pricing snapshot
Contact for pricing
Next step
Compare Marginal with similar tools before you shortlist it.
Compare this tool before you shortlist it
Review alternatives, pricing posture, and workflow fit side by side.
Marginal
Marginal provides cost observability for AI/LLM usage by logging every LLM request, computing its cost server-side against a daily-synced price catalog, and letting teams slice spend by any registered field such as customer, feature, or model. It offers SDKs (TypeScript, Python) and an HTTP API to send one event per LLM call; events are buffered and retried so instrumentation does not block application requests. The product includes an Explorer/dashboard that automatically charts spend by each dimension, highlights spikes, and records accepted/rejected events for debugging and audit.
Track every LLM call's cost and slice it by customer, feature, model — any field you define.
Own this listing?
Claim this page for a one-time $29 to add pricing, features, screenshots, verified owner details, and a clearly labeled 30-day category position after the profile is live.
Claim this listing for $29Key Features
Per-request cost tracking
Logs one event per LLM request and computes cost server-side using a daily-synced model price catalog so you don't need token math in your app.
Slice by arbitrary fields
Register fields such as customer or feature, then group and filter spend by those fields; unregistered keys are stripped and reported so dashboards stay clean.
LLM-aware pricing and frozen ingest prices
Cost is computed against the day's catalog rates with per-project overrides and prices are frozen at ingest; unknown models are flagged as unpriced and visible.
SDKs and HTTP API
Zero-dependency TypeScript and Python SDKs and a single JSON POST HTTP API for any language; SDKs buffer events and flush in the background with retries.
Explorer dashboard and saved views
Automatic insights charts for each dimension (model, provider, registered fields), plus the ability to save Explorer state (range, filters, group-by) as named views.
Event-level logging and debugging
Every API request is logged with outcomes (accepted, rejected, warnings, stripped keys, unpriced models) so you can debug integrations without guessing.
Integration recipes & coding-assistant support
Per-provider recipes and a marginalhq.com/llms.txt file to paste into coding assistants (Claude Code, Cursor) to automate wiring Marginal into your codebase.
Pricing
Current pricing details are not available from the vendor source.
Use Cases
Chargeback and customer-level billing
Group and attribute LLM spend by customer to identify top spenders and enable chargeback or allocation workflows.
Detect and investigate cost spikes
Automatically chart spend by model/provider/field to spot spikes and trace them to specific customers, features, or model changes.
Integration debugging and telemetry
Inspect per-event outcomes, rejected events, stripped keys, and unpriced models to diagnose integration issues and data problems.
Model and provider cost comparison
See spend and usage broken down by model and provider to inform cost optimization and model selection decisions.
Integrations
OpenAI
Examples show sending provider: "openai", model and usage from OpenAI responses to Marginal for cost computation.
Anthropic / Claude (and other providers)
Supports multiple providers and includes integration recipes; the site references provider examples like claude-sonnet-4 and coding-assistant recipes.
Coding assistants
Provides an llms.txt file that can be pasted into Claude Code, Cursor, or other coding assistants to automate integration.
Benefits
Limitations
Frequently Asked Questions
No verified FAQs are available.
Getting Started
- 1 Step 1: Create a project and register the fields you want to slice by (customer, feature, etc.).
- 2 Step 2: Install an SDK (npm install marginal-sdk or pip install marginal-sdk) or call the HTTP API directly with a POST to /v1/events.
- 3 Step 3: Add one track() call per LLM request including provider, model, usage, and fields; view automatic charts in Explorer and save useful views.
Support
docs
Site links to Docs and Quickstart guides for API shape, event conventions, and integrations.
demo / sales
Book a demo via the website ("Book a demo" link).
Contact via the provided email address [email protected].
API
https://marginalhq.com/docs
Compare Marginal with similar tools
See how it stacks up against alternatives
Related Tools
View all 140 →
DeepSQL
DeepSQL is a self‑hosted, 24/7 automated DBA and data engineering assistant that monitors database workloads, optimizes slow queries, and ships BI dashboards. It provides a built-in agent, CLI/MCP server, and Slack integration to let teams query, analyze, and put databases on autopilot while keeping data inside their VPC.
Tiny React charts that fit in a line of text, zero deps
microcharts is a zero-dependency React library of word-sized charts and sparklines (106 chart types) designed to fit inside a sentence, table cell, or streamed assistant reply. It emphasizes tiny bundle size, accessibility, theming, and AI-friendly machine-readable surfaces.
OrbitQuote
OrbitQuote is a procurement SaaS that extracts prices from supplier quotes (PDF, Excel, screenshots, WhatsApp messages), normalizes and cross-references items across suppliers, and builds the cheapest per-item purchase map so users can approve and export purchase orders quickly.
Voris
Voris is a privacy-first web analytics platform that tracks AI citations (ChatGPT, Perplexity, Gemini, Claude, Grok), ties those citations to site visits and conversions, and reports real-time campaign-level revenue and awareness — cookieless and EU-hosted.
Datarelax
Datarelax is an AI-driven data operations platform that unifies data modeling, ingestion, lineage, pipelines, and analytics into a single workspace, enabling engineers to design, deploy, and analyze governed data products from natural-language prompts.
Tekmatix
Tekmatix is an all-in-one, AI-powered business operating system and CRM that consolidates websites, funnels, email marketing, automation, AI 'employees' (bots), LMS for courses and memberships, payments and admin tools into a single SaaS platform for businesses of all sizes.
Askyourdatabase
AskYourDatabase is an AI-powered SQL chatbot and data analyst that lets users query, visualize, manage, and analyze databases using natural language. It offers a desktop app, embeddable chatbot, dashboard builder and supports many popular databases with enterprise security and on-premise deployment options.
Premium Alternatives
Usesaaskit
useSAASkit is a Next.js and React Native SaaS boilerplate that bundles AI-ready code (Vercel AI SDK), authentication, multi-organization support, admin tools, billing, marketing pages, analytics and i18n to help founders launch AI-powered SaaS and mobile apps quickly.
Documentpro
DocumentPro is an API-first document intelligence platform that extracts structured data from invoices, purchase orders, tax forms and other documents for embedding into software platforms; it emphasizes fast integration, AI-powered extraction, and maintenance-free production operation.
teamstation-ai
TeamStation AI is a Distributed Engineering Operating System (DEOS) and executive control surface for CTOs and CIOs that unifies talent-graph signal processing, neuro-psychometric math (Axiom Cortex), DEOS orchestration, Nearshore Control Plane governance, and delivery telemetry to govern LATAM nearshore engineering teams.
