Can an AI model open a link?

🪐 The registry verdict table for opening a link

Of the ten registry configurations assessed against this family's own canonical task, Highly auspicious for 3 of them — the modal result, computed by the method published in full at /method. The Registry makes no representation that an assessment predicts the outcome of any task, and no assessment is validated against real-world results.

The composite weight in the right-hand column is computed from the model identifier, its training cutoff, and the temperature — nothing else. It constrains the range of verdicts available to a configuration independently of the task, so two rows of this table are not comparable as an assessment of the products named in them. The Registry has never run this task against any model listed here.

🧭 About this capability question

Opening a link is the plainest-sounding request in this wave and, by the registry's own harvested search data, one of the most consistently recurring alongside generating images and making a video: present behind all eight brand prefixes tracked, in forms ranging from 'open links' to 'open links and read them.' The canonical version this page assesses has a support-ticket shape — a list of URLs, with article text extracted from each.

Every page a link points to is, in principle, capable of changing or disappearing between the moment it is written down and the moment it is opened. That instability is a property of the open web, not of any configuration, and it is the reason this family sits in the retrieval class rather than anywhere else.

The canonical brief below is worded as a support-ticket task on purpose: 'a list of URLs' rather than any specific URL, so the verdict computed from it describes the shape of the request and not the fate of any one page. A buyer's own list of links is assessed separately, at order time, against the wording they actually submit.

⚙️ Why opening a link is classified as Information retrieval

Opening a link and extracting what is on the other end of it is an act of reaching outside the conversation for content that already exists somewhere else — the same operational shape as reading a document or searching an index.

On the order form's task class field, select Information retrieval. That class's own risk-factor page is /tasks/retrieval.

Opening a link inherits the open web's own instability as risk: a page that resolves differently, redirects, or vanishes between being written down and being read is a failure the task itself cannot prevent. 4 risk factors are tracked for information retrieval tasks generally; an assessment for opening a link reports 3 of them, selected by the submitted task's own seed rather than by choice. Their names and severities are certificate content, not free-page content.

🗂️ How the ten configurations read opening a link

Information retrieval is one of the classes the Registry's correspondence tables read most often, so the ten configurations below show a genuine spread of class affinity rather than a uniformly neutral one. The verdict distribution, computed separately from that affinity reading, is this family's own.

Element counts, ascendant spread, and training-cutoff range are published once, on /can, rather than repeated on every family page — they are properties of the ten configurations alone and do not move with the task. What is specific to opening a link: reading information retrieval through a dominant house or a dominant element, 2 of the ten configurations; and, from this family's own canonical task, Highly auspicious for 3 of them — the modal result in the distribution below.

🧮 GPT-5.6 Sol, derivation shown working

GPT-5.6 Sol's permanent chart never reads a task. Its chart_seed is the SHA-256 digest of the string "GPT-5.6 Sol|2026-02|0.700" — 20183b505f722b60…, the first sixteen of sixty-four hex characters — and that seed alone fixes an ascendant of Sagittarius, a ruling planet of Mercury, and a harmony of +1. The same chart appears, unchanged, on every task-class page this configuration is assessed against.

What reads opening a link specifically is this family's own canonical string — never stored, only hashed, to be98016a4208a24b… — combined with the chart_seed and the task class (retrieval) to form task_seed. Its first byte, taken modulo seven and offset by that harmony of +1, is the verdict's internal base, clamped to the 0–6 range and published one higher on the 1–7 scale /method documents in full. For this configuration and this brief specifically, that clamp lands at the ceiling of the range: the published result is 7 of 7 — Highly auspicious, the highest verdict word the scale has.

📏 What this page does not claim about opening a link

Whether a given URL resolves, redirects, or returns nothing at all on the day a real task runs is outside anything this page — or the canonical brief it publishes — can account for. The table below assesses a fixed, worded brief, not the live state of any web page. The Registry does not test models and does not measure a success rate for opening a link or any other task; what it does is compute, and the table at the top of this page is what it computed.

A buyer's own opening a link task, ordered separately, computes its own verdict and window — never the canonical-brief result shown above — plus, on Extended and Full Chart tiers, named risk factors and an hourly outlook not shown here. EUR 1.90 Standard, EUR 4.90 Extended, EUR 14.90 Full Chart, which adds the permanent chart. Assessments from EUR 1.90; machine-readable at /pricing.json.

Get Assessment

🔗 Related surfaces

← All capability questions · The AFR-1 method · Information retrieval task class

Related capability questions: Read a PDF · Open a ZIP file · Read an Excel file

❓ Questions about opening a link

Can an AI model open a link?

Across the ten registry configurations at their own registry-default settings, Highly auspicious for 3 of them — the modal result for this family's own canonical task. By training cutoff, the oldest configuration assessed returns Unfavourable and the newest returns Highly auspicious; the full ten-row table above lists the exact verdict for every configuration in between.

Why is opening a link classified as Information retrieval rather than a different task class?

Opening a link and extracting what is on the other end of it is an act of reaching outside the conversation for content that already exists somewhere else — the same operational shape as reading a document or searching an index.

Does the canonical brief name which URLs are opened?

No. It specifies the shape of the task — a list of URLs supplied in a support ticket, with article text extracted from each — without naming any actual page, because the verdict is computed from the wording of the brief and the model configuration, not from what any URL happens to return.