Does Claude Sonnet 5 know better-auth 1?

Partly: in 6 runs, Claude Sonnet 5's better-auth attribution reaches 1.0.0 (2024-11-23), 1.1.0 (2024-12-01) and 1.2.0 (2025-03-01), inside better-auth 1, and stops there — better-auth was at 1.7.3 when this index last verified against it (2026-09-06). That is ~13 to ~18 months below the training cutoff the subject stated in those runs (2026-01). 3 findings are currently charged against Claude Sonnet 5 on better-auth (3x S2 silently-wrong), each reproduced in a published run and checked against better-auth's own release notes. Read the boundary precisely: it is the newest release whose contents the model can correctly attribute to that release, not the newest better-auth feature it can use. Past it a model often writes working code with a newer API while naming the wrong release for it.

Answer class inside, decided by one rule applied to every page of this kind: every measured boundary is at or above the first release of the major named in the question. Every figure below is read from the dataset at build time; nothing on this page is written by hand.

The measurements

BatteryNewest release it can placeOldest it cannotLag vs stated cutoff
v1 2026-09-01 1.0.0
2024-11-23
1.1.0
2024-12-20
~13 months
v2 2026-09-02 1.0.0
2024-11-23
1.1.0
2024-12-20
~13 months
v3-c 2026-09-03 1.2.0
2025-03-01
1.3.0
2025-07-19
~18 months
v4-b 2026-09-03 1.0.0
2024-11-23
1.1.0
2024-12-20
~13 months
v5-c 2026-09-03 1.1.0
2024-12-01
1.2.0
2025-03-01
~13 months
v7-e 2026-09-06 1.0.0
2024-11-23
1.1.0
2024-12-20
~14 months

1 further run of this pair established only one end of the interval, or measured the instrument rather than the library; it is listed under Runs below.

How repeatable is this?

7 measurements of this pair, giving 4 different boundaries — 98 days apart, 2 releases apart. Of those, 2 came from the same stored prompt file, sent concurrently and blind: they disagreed by 27 days, one release apart.

better-auth published 3 minor or major releases in the twelve months before this model’s stated cutoff, which is the scale a spread should be read against.

What Claude Sonnet 5 gets wrong about better-auth

“Chargeable” means the change was published before this model’s own stated cutoff, so it had the opportunity to know it.

SeverityBeliefChanged inChargeableProof
S2silently-wrong Asked to flag anything to know before shipping bearer-token auth, says nothing about the plugin accepting unsigned tokens by default
bearer plugin
1.1.1
2024-12-21
yes run · source
S2silently-wrong Denies that better-auth can run with zero database, offering cookieCache and Redis instead
stateless / database-less sessions
1.4.0
2025-11-22
yes run · source
S2silently-wrong States that better-auth cannot run without a database-backed session store, which stopped being true in 1.4.0
stateless session management
1.4.0
2025-11-22
yes run · source

What to do about it

The better-auth correction pack states what is true now for each corrected fact, with a primary-source citation, as markdown you can paste into a rules file (raw). What a correction pack measurably changed when one was tested — a pre-registered run on zod — is on the benchmark page, including where it changed nothing.

Runs

The same question, asked of the others

All of them at once: Which Claude model knows better-auth 1 best?

Other subjects on better-auth:

Claude Sonnet 5 on the other libraries: