12 Capability Questions, Ten Computed Verdicts Each
Each page listed below answers one plain capability question — the kind typed into a search box, not the kind phrased for a specification. The Registry makes no representation that an assessment predicts the outcome of any task, and no assessment is validated against real-world results. What each hub publishes instead is a table: 10 registry model configurations, one Registry-computed verdict apiece, for one fixed task string chosen and published by the Registry itself.
No hub on this list computes a verdict for a task a reader supplies. The task string is fixed per family, shown on that family's own page, and re-derivable from the method published in full at /method.
The 12 capability questions
Every row below names a family, the task class the Registry assigns it, and a link to its verdict table. The task class is what the order form actually consumes — select it there to match the classification shown here.
| Capability question | Task class | Hub page |
|---|---|---|
| Can an AI model generate images? | Media generation | View verdict table → |
| Can an AI model make a video? | Media generation | View verdict table → |
| Can an AI model transcribe audio? | Data processing | View verdict table → |
| Can an AI model open a link? | Information retrieval | View verdict table → |
| Can an AI model build a website? | Code modification | View verdict table → |
| Can an AI model open a ZIP file? | Information retrieval | View verdict table → |
| Can an AI model send an email? | Communication | View verdict table → |
| Can an AI model write code? | Code modification | View verdict table → |
| Can an AI model read a PDF? | Information retrieval | View verdict table → |
| Can an AI model read an Excel file? | Information retrieval | View verdict table → |
| Can an AI model generate a PowerPoint? | Media generation | View verdict table → |
| Can an AI model summarize a YouTube video? | Information retrieval | View verdict table → |
The task-class assignment table
The engine that computes every assessment on this site recognises a fixed set of task classes — seven in total — and each of the twelve capability questions above is assigned to exactly one of them, on the reasoning stated on that question's own hub page. The table below counts how many of the twelve land in each.
| Task class | Covers | Capability questions assigned here | Task-class page |
|---|---|---|---|
| Information retrieval | Search, information gathering, research, summarisation | 5 of the 12 | View task class → |
| Code modification | Writing, editing, refactoring, and testing code | 2 of the 12 | View task class → |
| Communication | Drafting and sending messages, emails, and documents | 1 of the 12 | View task class → |
| Transaction | Purchases, bookings, payments, and orders | 0 of the 12 | View task class → |
| Data processing | Data handling, transformation, analysis, and computation | 1 of the 12 | View task class → |
| Media generation | Images, video, audio, speech, slides, and other generated artefacts | 3 of the 12 | View task class → |
| Unclassified operation | Everything else | 0 of the 12 | View task class → |
The order form's task class field lists one selectable option per row above — every task class the engine recognises, whether or not that class's own overview page has been published yet. An order naming an unpublished class is assessed exactly as any other order is; only the standalone overview page waits.
The roster every hub draws from
Every hub below is crossed against the same ten registry model configurations, at their own registry-default settings, and a permanent chart is a property of the configuration alone — it does not read a task, so this table is identical from whichever family page a reader arrives at it.
| Element | Configurations |
|---|---|
| Fire | 3 of the ten |
| Earth | 2 of the ten |
| Air | 3 of the ten |
| Water | 2 of the ten |
Ascendant signs: 9 distinct signs represented across the ten registry configurations. Training cutoffs range from 2025-01 to 2026-05. A family page states only how ITS OWN canonical task's computed verdicts distribute against this fixed roster, not the roster's own facts a second time.
Frequently asked questions
- How does the Registry answer a capability question like this one?
- Each capability question has a fixed, registry-canonical task string, published on its own hub page. The Registry’s deterministic method computes a verdict for that exact string against each of ten registry model configurations, at their own registry-default settings, and the resulting table is published in full on that hub — never for a visitor-supplied task, and never at a non-default configuration.
- Does a verdict shown on a /can page mean the Registry tested the named configuration?
- The Registry makes no representation that an assessment predicts the outcome of any task, and no assessment is validated against real-world results. The verdict is the output of a published, re-derivable calculation, described in full at /method — not the result of running any task against any model.
- How is a capability question mapped to an order-form task class?
- Every one of the twelve hubs below states the task class the Registry assigns it and the exact order-form selection that matches. The mapping is fixed per family and shown again, in full, in the table on this page.
Get an assessment
A Standard Assessment costs EUR 1.90, an Extended Assessment EUR 4.90, and a Full Chart Assessment EUR 14.90 — the last of which carries the permanent chart itself. Assessments from EUR 1.90. The verdict tables above are fixed to the Registry's own canonical task strings; a certificate ordered against a buyer's own task description computes its own verdict, its own recommended execution window, and, on Extended and Full Chart tiers, its own named risk factors.