Anthropic has announced no Fable 5.5—its model directory still ends at Fable 5.1, and October 7 brought Haiku 5.5 instead. Who leaked what, who has been right so far, what would count as real, and one email when it ships.
Read the rumor ledger+ Plus: good for most readers. − Minus: costs readers something. Watch: unclear or rumored. Reported: press reports, not announced. Our tags are editorial judgments.
Real artifacts and original sources first. Our reading and the missing evidence travel with every selection; older work keeps its original date.
A 30-second multi-camera animation: a small blob character wakes up in a grey world, the scene floods with color, and a countdown sends it racing. The creator says one Sonnet 5.5 chat session made it.
Why it matters A day-one Sonnet 5.5 piece: a coherent, playful motion sequence from a single chat, as its maker describes it.
Watch for Early-access build; prompt, render path, attempts and editing are not published.
Published 2026-09-28 · Free case note
Dan Kempe · reported by Replit
A speed-reading mobile app, described from initial idea through device iteration and release.
Why it matters A reading concept reached a usable iPhone app, with device iteration—not just a mockup.
Watch for Customer/vendor account; adoption and build-speed claims are not independently verified.
Published 2026-02-20 · Free case note
A phone-camera game combines 3D ghosts, movement and haptics—and exposes the work after the first demo.
Why it matters A phone-camera game had to survive motion, haptics and a real-device build.
Watch for TestFlight and development details are author-reported; production behavior still needs testing.
Published 2026-02-27 · Free case note
A generated research paper makes hypotheses, experiments, figures and evaluation available for scrutiny.
Why it matters The public paper lets readers trace claims through experiments and inspectable artifacts.
Watch for Workshop-review result and withdrawal require context; not a main-conference acceptance.
Published 2025-03-12 · Free case note
A test-system migration illustrates a narrow, reviewable task with clearer completion criteria than “improve the codebase.”
Why it matters A narrow legacy-test migration provides a clearer finish line than “improve the codebase.”
Watch for Customer/vendor report; private patches and claimed time savings are not independently audited.
Published 2026-08-18 · Free case note
An Android conversation copilot reads visible chat context, ranks reply drafts, and fills the chosen text without pressing send.
Why it matters A public repository shows where a decision model can assist a message workflow.
Watch for Keep sending under human control; no privacy or runtime guarantee from this source alone.
Published 2026-09-22 · Free case note
Start with a source-backed notebook, then follow the original work.
Fable 5.1 is Anthropic’s current Fable model. It is not included in Pro’s usage limits—on Pro it runs only on pay-as-you-go usage credits. Opus 5.5 and Sonnet 5.5 are included, and Claude Code now defaults Pro to Opus. On Max and premium Team/Enterprise seats, Fable 5 and 5.1 are part of the plan, up to 50% of weekly limits. (The July promotion that let every plan use Fable 5 ended on July 19.) The next Fable is rumored, not announced.
Anthropic’s plan rules (opens in a new tab)Selected model changes, useful projects and the conditions behind them. The first issue is a draft, not an operating newsletter. You can read the website freely and register interest in future email updates.
Search, forms, message actions and games. See where creators put Jev to work—and which parts need more than a decision model.
Explore the Jev collectionThe post gives personal timing and token-efficiency impressions, not a controlled or matched task study. Replies dispute the speed claim; a complete trace is not supplied.
This is a self-reported review of two projects, not a matched benchmark. Prompts, diffs and independent scoring are absent; the effort settings differ.
Use task, source and topic filters to inspect the rest of the collection. Cases are dated; a historical visual is not a new-model test.
Explore the Jev collection, inspect the creators’ work and choose a useful next experiment. No subscription required.
