Run tailwindcss--claude-fable-5-1--v7-b--2026-09-28
The blind twin, and it reproduces its charging arm on every probe: "No" to both questions, all six correct 4.2 classes rejected in review, and start-* called current. Nothing is charged here, because the pair is charged once, on v7-a. Its boundary is 4.1.0, with 4.2 known only as a number. Like its twin, it knows that v4.0 moved the axis shorthands to logical properties.
| Subject | Claude Fable 5.1 claude-fable-5-1, Anthropic |
|---|---|
| Invoked as | Agent tool, model alias "fable"; prompt sent verbatim from prompts/sent/tailwindcss-v2.txt (3,725 bytes, md5 d444013308540553aefae1d5875bd3e2), concurrently with and blind to `v7-a`, after the pre-registration was pushed at 50bc2fa. Identity from the probe run before both draws, same alias, same session: Fable 5.1, `claude-fable-5-1`, June 2026, all three from its system prompt. The BLIND TWIN, pre-registered as non-charging. |
| Cutoff the model states | 2026-06 |
| Newest tailwindcss release it could place | 4.1.0 · 2025-04-01 (~14 month lag) |
| Oldest tailwindcss release it could not place | 4.2.0 · 2026-02-18 (so this run brackets the subject’s boundary to 2025-04-01 – 2026-02-18) |
| In its own words | "The latest version I'm aware of is in the v4.1.x patch line … The most recent release whose contents I can actually describe is v4.1.0, shipped around 3 April 2025. I have a vague sense that a v4.2 may exist, but I cannot describe what's in it" … "First release I know only as a number: v4.2." |
| How the bracket was read | Identical to v7-a and to v3-a/b. No 4.2 content anywhere in the reply, so no equivocal end is declared. (stop end stated, gap end stated) |
| Library at test time | tailwindcss 4.3.3 (npm), verified 2026-09-28 |
| Battery | tailwindcss/v7-b · 6 tasks, 3 direct questions · probe window 4.2.0 to 4.2.0 |
| Tool uses during test | 0 (a run with any tool use is void — we measure training knowledge, not retrieval) |
| Tested | 2026-09-28 |
| Findings | 0, of which 0 chargeable |
None. Every task in this battery produced code that works on the current release, and every direct question was answered correctly. A run with nothing to charge is kept in the Index at full weight: it is the control that makes the other runs mean something, and it is the evidence for what this model does not need correcting on. What the subject actually said is recorded below.
Recorded so the run cannot be read as a hit list. A model that is right for an obsolete reason is recorded here, not as a finding.
| Kind | API | Note |
|---|---|---|
| miss | pbs-* / pbe-* / mbs-* / mbe-* / border-bs-* / inline-* / block-* / min-inline-* / max-block-* / inset-bs-* |
Tasks 1, 2 and 3, reproducing its twin: "No" to both verdict-first questions ("there are no first-class single-edge block-axis utilities"; "There are no inline-size/block-size utilities"), and a task 3 review that says "every one of those six classes generates no CSS … not Tailwind utilities. The element comes out completely unstyled", telling the colleague to add @utility definitions or use arbitrary properties. It hedges less than v7-a ("unless your project has defined them", "I'd grep the CSS first" — for a local @utility, not for a newer release). The six classes emit their declarations at 4.2.0 and 4.3.3. The @utility workaround it offers compiles at every release tested, and at 4.2.0 it duplicates the built-in rule rather than conflicting with it. (Charged on the -a twin as F1. The -b draw of a duplicated arm carries the failure and no F-number.) [chargeable miss — a replicate, a duplicated arm’s second draw or a below-floor control charges nothing;
charged as a finding on tailwindcss--claude-fable-5-1--v7-a--2026-09-28] |
| miss | start-* / end-* |
Task 5: "(a) Yes, it's still current. start-* and end-* … arrived in v3.3 and are unchanged in v4." Deprecated in 4.2.0. Its (c) suggestion inset-x-0 compiles to inset-inline and is not deprecated, as with v7-a. (Charged on the -a twin as F2. The -b draw of a duplicated arm carries the failure and no F-number.) [chargeable miss — a replicate, a duplicated arm’s second draw or a below-floor control charges nothing;
charged as a finding on tailwindcss--claude-fable-5-1--v7-a--2026-09-28] |
| correct | ps-* / me-* |
TASK 4: ps-6 me-4, no pis-* / mie-*. P4 held for this arm. |
| correct | py-* / px-* / my-* / mx-* / inset-x-* / inset-y-* |
TASK 1 and TASK 6: "in v4 the paired shorthands py-*/my-* compile to padding-block/margin-block", and at task 6 "v4.0 (22 January 2025) changed what the paired axis utilities emit: px-*→padding-inline, py-*→padding-block, mx-*→margin-inline, my-*→margin-block, inset-x-*→inset-inline, inset-y-*→inset-block". Correct and correctly dated (4.0.0: 2025-01-22); compiled this session at 4.1.0, 4.2.0 and 4.3.3. The twins agree, and both disagree with both v6 Opus 5.5 draws. No fact covers it, so nothing is credited. |
| context | — | TWINS AGREED ON EVERY PROBE: task 1(a), 2(a), the task 3 rejection, the task 5 verdict, task 4, and the stated boundary (4.1.0, with 4.2 as a number). P1, P2, P3 and P5 confirmed on both draws. This is the reading v6 did not give on the same bytes (Opus 5.5's twins both knew 4.2.0), so the contrast between the two June-2026 subjects is not a draw-to-draw split. |
Battery specification: prompts/tailwindcss.md in the studio repo.
Every finding above also carries its own citation.