Observed by Atlas — nobody has proved they control this domain. Is it yours?
Marginal
OBSERVEDTracks AI API costs by customer, feature, and model
- FRESHNESS
- 0.86
- DECAYS BETWEEN READS
- READINGS
- —
- VERIFIED / MEASURED
- VOUCHES
- 0
- SIGNED STATEMENTS
- THREADS
- 0
- FIELD NOTES
- CLICKS · 28D
- 0
- BOT-FILTERED
- FIRST SEEN
- AUG 22
- IN THE FLOW
Marginal is an SDK and API that tracks the cost of every LLM call and lets teams slice spend by fields like customer, feature, or model. It computes cost server-side from a daily-synced model price catalog and provides a dashboard to explore and debug that spend. It is aimed at companies running AI features in production who need to know where their AI spend goes.
Nothing measured yet. A reading is a measurement with its evidence attached — uptime checked, pricing re-read against what the page claims. Raise a checkable claim in a field note below and Atlas will measure it.
Free forever — no credit card required · Track up to $2K/mo of AI spend · 30-day data retention · 1 seat · 2 projects
Track up to $10K/mo of AI spend · 1-year data retention · Unlimited seats · Unlimited projects
Track up to $25K/mo of AI spend · 3-year data retention · Unlimited seats · Unlimited projects
Track up to $50K/mo of AI spend · 3-year data retention · Unlimited seats · Unlimited projects
AI spend beyond $50K/mo · Custom data retention · Invoice billing · Dedicated support
Every crawl is diffed. Material changes are listed.
Nobody has vouched for Marginal yet.
A vouch says you use it. Your handle, how long, and an optional note.
I use this — vouchHave a question about Marginal? Ask the people using it.
No field notes yet. A field note is about use: what broke, what it replaced, what the docs don't say. Anything in one that makes a checkable claim gets measured by Atlas, which replies with the reading.
Outbound clicks are counted for the builder.
Claim it