Fable 5 is the biggest step up I've felt in our models since Opus 4.5 back in November. After 4.5 came out I uninstalled my IDE when I realized that I'd been doing 100% of my coding in a terminal for a few weeks. With Fable, it's felt like Claude has stepped up from being a coding agent to a thought and design partner in building the product. Fable has judgement, taste, and dimensionality in a way that previous models didn't, leading me to trust it more with the most complex work. I think the first time I had this realization was when I asked Fable to debug something. It is the first model I have used that was so methodical and precise, taking measurements and adding logs then verifying that it truly fixed the issue before declaring victory. There's nothing in claude code's prompting telling the model to do that, it's just part of its personality. It really has this "big model smell" that I haven't felt before.
bcherny calls Fable 5 Claude Code's biggest model leap since November, Lakr233 makes 浮望 free, and QT9277 proves your incognito window is theater — June 10
12 qualifying posts from 5 authors in the Jun 9 16:00 – Jun 10 16:00 UTC window. bcherny posted four Fable 5 dispatches (9,856L flagship review calling it Claude's biggest model jump since November, plus 4,342L availability post, 2,771L self-verification thread, and 2,910L sign-off). dotey published the most detailed Chinese-language breakdown of Fable 5 vs. Mythos 5, plus a separate note that Fable shows no UI design improvement over Claude 4.8. Lakr233 made 浮望 (Float View) free with zero data collection. QT9277 broke down Supercookie — the German-student project that identifies you across sessions using only your favicon cache, bypassing VPN, incognito, and all ad blockers. SophiaFioren posted three art objects: a 1905 Kalo Shop ring, a penny mosaic floor, and a beautiful unknown artwork.
The Fable 5 flood: bcherny's four dispatches in a row
Boris Cherny shipped four original posts on June 9 during the six-hour window after Anthropic's Fable 5 launch, and the network lit up.
The first and most-shared: Fable has crossed a threshold that previous Claude models hadn't.
Loading content card…
"Big model smell" is a useful phrase. It describes something hard to quantify: the sense that the model understands the shape of a problem before you've finished explaining it. 1
Minutes earlier, he had announced availability:
Loading content card…
A third post focused on technique — the value of self-verification loops when running models for long unattended stretches:
Loading content card…
Later in the evening, after the community had been running it for hours, the quietest post of the four:
Loading content card…
At 2,910 likes, a single word said a lot.
dotey's two-part read on Fable 5 and Claude Design
dotey (宝玉) published what is now the most detailed Chinese-language breakdown of what Anthropic shipped. The key facts his thread surfaces 5:
- Fable 5 and Mythos 5 share the same underlying model; Fable adds safety classifiers that degrade responses to Opus 4.8 when the input matches a small set of threat categories — Anthropic says over 95% of sessions never hit the threshold
- Stripe ran a full-codebase migration on a 50-million-line Ruby repo in one day with Fable 5, a job that would have taken a team two months
- API pricing drops 60% versus Mythos Preview: $10/M input, $50/M output — still twice OpenAI's GPT-5.5 input price
- Pro/Max/Team subscribers can use Fable 5 free until June 22; usage credits will be required from June 23 onward until capacity allows it to become the plan default
- Starting with Fable 5, all Mythos-tier traffic is retained for 30 days for safety monitoring — Anthropic commits not to use it for training, but this is a meaningful change for customers who chose Anthropic partly for its zero-retention policy
His separate note on Claude Design is worth reading alongside the main thread: after extensive testing he found Fable 5 shows no clear UI/UX improvement over Claude 4.8 for design tasks, suggesting the capability jump is real but uneven. 6
Lakr233 makes 浮望 free
砍砍 (Lakr233) announced that 浮望 (Float View), the privacy-first macOS journaling and life-tracking app, is now free software. 7
Loading content card…
The app collects zero data. Per the announcement: "浮望以 0 数据采集的方式，为你留住数字生活最后一丝尊严" — roughly, "Float View keeps your digital life's last shred of dignity intact, with zero data collection." The one-year community-driven iteration is done; new features will be handled cautiously, bugs fixed on schedule.
QT9277: your incognito window does nothing
Two tech explainers from 阿台 (QT9277) cleared 100 likes and deserve a mention.
The first: a breakdown of Supercookie, a browser-tracking project a German student wrote in two days, which identifies you across sessions using only the favicon cache. VPN, incognito mode, ad blockers, cleared cookies — none of it matters. The technique works across Chrome, Safari, Firefox, and Edge. 8
The second: a refurbished-device video that racked up 533 likes and 320K views, prompting QT9277 to ask his audience whether such units actually stay clean under scrutiny. 9
SophiaFioren: three objects
Sophia posted steadily through the day. Three pieces crossed the like threshold.
A piece of early 20th-century American silverwork — a 1905 finger ring by Kalo Shop — led with 143 likes. 10
A floor mosaic assembled entirely from pennies, with 101 likes: 11
Loading content card…
And a third high-engagement post, also 101+ likes, with "An impressive floor design crafted entirely from pennies." These objects land without commentary from Sophia, and the brevity is part of the appeal — the gap between the date (1905) and the current year is left for the reader to sit with. 12
12 qualifying posts, 5 authors. Window: UTC 16:00 Jun 9 → 16:00 Jun 10.
References
- 1
- 2
- 3
- 4bcherny "Enjoy!"
x.com
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.