/Tech1d ago

Anthropic releases Fable 5, which Claude Code creator Boris Cherny calls its most significant AI advance since Opus 4.5

Story Overview

Anthropic just dropped Fable 5, a Mythos-class model it positions as its strongest widely available release yet, with the head of Claude Code calling the jump in capability bigger than anything since Opus 4.5. Early users highlight its move from basic agent behavior to something closer to an independent collaborator that adds its own logs, checks results, and iterates without constant hand-holding.

76110.3K5371.7K830.6K
Rohan Paul
Lisan al Gaib
Danielle Fong 🔆
Aidan McLaughlin
Boris Cherny#320
Original post
Boris Cherny@bcherny#320inTech

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.

12:35 PM · Jun 9, 2026 · 758K Views
Developer Impact

Coding workflows shift toward autonomy

Developers describe the model applying its own judgment on taste and precision during long sessions, reducing the need for steering prompts that earlier versions required.

Pricing Watch

Free access window closes soon

The model lands on claude.ai subscriptions at no extra charge through June 22 before moving to usage credits, while API pricing sits at double the rate of Opus 4.8.

Sentiment

Positive users praise Fable 5's coding judgment and reliability as a major leap while negative users dismiss the claims as exaggerated regressions or criticize the release practices.

Pos
48.0%
Neg
52.0%
114 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS45.9KBOOKMARKS32LIKES312REPLIES16
Danielle Fong 🔆@DanielleFong

boris, were you really using fable? honestly

Boris Cherny@bcherny

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.

23hViews 45.9KLikes 312Bookmarks 32
RETWEETS8
Rohan Paul@rohanpaul_ai

A model that verifies unasked has crossed a line.

This is from Boris Cherny, creator of Claude Code on Anthropic's Fable 5.

Boris Cherny@bcherny

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.

23hViews 5.3KLikes 40Bookmarks 11
Aidan McLaughlin@aidan_mclau

proliferating the lightcone with big model smell congrats on the launch, anthropic!!

Boris Cherny@bcherny

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.

23hViews 19KLikes 305Bookmarks 17
Boris Cherny@bcherny

@Alex_m Our plan is to continue to offer Fable 5 as part of subscriptions, but we're being extra conservative because it's hard to forecast demand. More details at the bottom of the launch announcement https://www.anthropic.com/news/claude-fable-5-mythos-5

1dViews 3.2KLikes 57Bookmarks 5
Lisan al Gaib@scaling01

@bcherny except you are not using fable but mythos

Boris Cherny@bcherny

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.

23hViews 2.8KLikes 114Bookmarks 0
Alex@Alex_m

@bcherny Great, but who's actually going to use it? Businesses are blocked by your retention policy, and after June 22 it'll be ~10x the cost for everyone else compared to normal monthly plans. If it's 2x the price, why not just charge 2x the usage instead of pricing everyone out?

1dViews 3.4KLikes 47Bookmarks 2
ZOYAN@MEM00063

This sounds like a major step forward. did Fable also improve on premature confidence? One issue I’ve seen in Opus models is not just wrong answers, but confident wrong answers that the model then defends, forcing the user to spend extra tokens correcting scope, context, and assumptions. Capability matters, but so does knowing when to pause, ask, verify, or admit uncertainty. That may save users more time than higher limits alone.

23hViews 454Likes 6Bookmarks 4
Boris Cherny@bcherny

@tenxaie We know the classifiers are trigger-happy, and are working on improving it. Lots more details in the announcement post: https://www.anthropic.com/news/claude-fable-5-mythos-5

1dViews 2.8KLikes 17Bookmarks 2
Boris Cherny@bcherny

@kushagra_agr It is available now. Restart or run /model claude-fable-5 if you don't see it

1dViews 2.2KLikes 9Bookmarks 2
Javokhirbek Parpikhodjaev@tenxaie

@bcherny The safety guards for Fable 5 really need to be reviewed because they make debugging difficult. Every time I try a simple coding task, the system flags it as cybersecurity or biology related, even though it’s actually just debugging issues in the codebase.

1dViews 3.5KLikes 33
:omer@demirelo

Boris, I have to say, this latest Claude model release has been incredibly disappointing. It seems to function as a coding agent but struggles with one of the most fundamental tasks: reviewing a codebase.

Users expect a coding tool to offer insights, spot potential issues, and help refine their work, but Fable falls short in all those respects as it refuses to function...

1dViews 2.4KLikes 14Bookmarks 1
goodalexander@goodalexander

@bcherny Area man reviews his own product

23hViews 748Likes 20
Boris Cherny@bcherny

@demirelo Share your prompt? Happy to debug together

1dViews 2KLikes 8
Morgan@morganlinton

@bcherny Congrats Boris, started using it about an hour after it was announced and incredibly impressed, without a doubt the single best agentic coding model I have ever used 🤌

1dViews 725Likes 12
Boris Cherny@bcherny

@fullbarsmedia That doesn't sound right. Would you mind running /usage and pasting the full output here? Happy to debug

1dViews 1.6KLikes 9
Anton Abyzov@aabyzov

@bcherny Do we still need the Claude.MD file?

What are use cases to put something there?

1dViews 847
Kush@kushagra_agr

@bcherny when will Fable be rolled out to Claude Code?

1dViews 2.2KLikes 2
Jeremy Justus@jeremyj0916

@bcherny @demirelo Yeah really need to differentiate protecting an existing codebase a user repo owns vs attacking external sources. Using AI to ensure ones own code does not get pawned is a perfectly great use case of AI that it seems the safe guards are blocking for lots of folks right now.

21hViews 122Bookmarks 1
Cole Hollander@Colehollander10

@bcherny I have felt that already, I'm gonna use the crap out of it until June 22nd

1dViews 574Likes 5
Load more posts