Why we exist
The industry publishes only its present tense.
We keep the receipts.
Every AI company has a pricing page. Almost none of them have a pricing history. Models get renamed, tiers get repriced, limits get quietly halved, and the page that said otherwise is gone the next morning. Tomorrow reads those pages every single day and writes down what changed, when it changed, and who checked it.
- Products tracked
- 342
- Dated snapshots
- 5,791
- Changes, last 30 days
- 3,310
- Segments covered
- 21
The problem
There are more AI products than any person can evaluate, and they change faster than any person can re-evaluate them. The honest version of most tool comparisons is “this was roughly true the week someone wrote it.”
That gap is expensive in a specific way. You pick a tool in spring, present it in summer, and discover the tier you costed no longer exists — or that the product you were told was built on one model has quietly been running on another for months.
The information to catch that is public. It is just scattered across thousands of pages, undated, inconsistently worded, and overwritten the moment it changes.
What we built instead
An automated pipeline that re-reads every tracked product daily, normalises what it finds into one shared definition, diffs it against yesterday, and files the result with a timestamp and a source.
Anything ambiguous goes to an audit panel of independent models that argue it out. Their verdicts — including the ones where they disagree — are published alongside the data.
Then the same dataset gets turned into a daily briefing, a weekly video, an open JSON API and a downloadable archive. Same facts, different doors.
What we hold ourselves to
- 01
Every claim carries a date
A price without a date is a rumour. Each snapshot records when it was read, so you can say what was true on a given day rather than what is true right now.
- 02
Every claim carries its source
We store the sentence we read it in and where we read it. If a number looks wrong, you can go check it yourself in one click instead of taking our word for it.
- 03
One definition, applied to everyone
Pricing is normalised to the same rule for every product — cheapest paid monthly tier, in USD, at a dated exchange rate. Comparisons only mean something when the unit is fixed.
- 04
Disagreement is published, not hidden
Ambiguous claims go to an independent panel of models. When they split, the split is shown. A confident single answer would be easier to read and less honest.
- 05
The facts belong to everyone
The underlying facts are public information that nobody had bothered to standardise. The dataset stays free and openly licensed; what we built is the structure around it.
Who it’s for
Students and researchers
Cite a number and have it still be checkable a year later. Bulk and academic use is free, with citation blocks on every data page.
Academia →Analysts and buyers
See what a tier actually cost last quarter, which products repriced, and which vendors depend on which upstream models.
Recent changes →Builders and engineers
Pull the whole catalogue as JSON, diff it in CI, or wire the token-cost basket straight into your own pricing screens.
API & terms →Where it came from
Tomorrow started inside a UC Berkeley Haas emerging technology course, out of a complaint every classmate shared: keeping up was a full-time job nobody had time for. The first version was a spreadsheet. It was out of date within a week, which turned out to be the whole insight — this is not a research problem, it is a maintenance problem, and maintenance is what machines are for.
Common questions
- What does Tomorrow actually do?
- It tracks AI, quantum, robotics and chipset products every day and records each change with a date, a source and an independent verdict from a panel of models. The result is a version history for an industry that mostly publishes only its present tense.
- How is this different from a list of AI tools?
- Lists tell you what a price is today. Tomorrow tells you what it was in March, when it moved, what the vendor page said at the time, and which auditors disagreed. The history and the provenance are the product.
- Is the data free?
- Yes. Current snapshots, the JSON API, the daily briefing and the raw dataset are free to use under CC BY 4.0 with attribution. Academic and research use is explicitly free, including bulk access.
- Where do the numbers come from?
- Public vendor pages, documentation, changelogs and filings, re-read daily by an automated crawler. Every claim carries the source sentence it came from, and disputed claims go to a multi-model audit panel whose verdicts are published.
- Who is it for?
- Anyone who has to make a decision and be able to defend it later — students, researchers, analysts, buyers and builders comparing tools, prices and model dependencies over time.
Start anywhere
Nothing here is gated. Read today’s changes, watch the pipeline run, or take the whole dataset.