/AI2h ago

Anthropic resets usage limits for its new Fable model, advising developers to use broader prompts and higher effort

Legacy CLAUDE.md instructions can restrict the model's judgment.

971K5044391.4K
Chubby♨️
Alex Volkov
Alex Albert#440
Original post
Alex Albert@alexalbert__#440inAI

We've reset usage limits across our products!

For those just starting to test Fable, here's four tips for using it more effectively: 1. Give it bigger, more ambitious tasks than what previous models could handle. 2. Use xhigh/high effort as your default for best performance, med for faster interactive sessions. 3. Rework your skills and CLAUDE.mds. Instructions written for prior models anchor Fable to stale patterns, let it use its own judgment first. 4. Move from providing tasks to providing objectives. Describe what done looks like and how to verify it, then let Fable find the path (/loop and /goal are built for this)

3:00 PM · Jun 9, 2026 · 89.3K Views
Sentiment

Positive users appreciate the Fable 5 launch and usage reset for enabling bigger tasks, while negative users criticize the ongoing limits, false flags, and restrictions making it unusable.

Pos
19.1%
Neg
80.9%
45 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS6.4KLIKES87RETWEETS4
Chubby♨️@kimmonismus

yay weekly reset, thanks!

Alex Albert@alexalbert__

We've reset usage limits across our products!

For those just starting to test Fable, here's four tips for using it more effectively: 1. Give it bigger, more ambitious tasks than what previous models could handle. 2. Use xhigh/high effort as your default for best performance, med for faster interactive sessions. 3. Rework your skills and CLAUDE.mds. Instructions written for prior models anchor Fable to stale patterns, let it use its own judgment first. 4. Move from providing tasks to providing objectives. Describe what done looks like and how to verify it, then let Fable find the path (/loop and /goal are built for this)

1hViews 6.4KLikes 87Bookmarks 1
BOOKMARKS1REPLIES4
Alex Volkov@altryne

Lol this is my (MAX) usage post reset, on one dynamic workflow. 😅

Alex Albert@alexalbert__

We've reset usage limits across our products!

For those just starting to test Fable, here's four tips for using it more effectively: 1. Give it bigger, more ambitious tasks than what previous models could handle. 2. Use xhigh/high effort as your default for best performance, med for faster interactive sessions. 3. Rework your skills and CLAUDE.mds. Instructions written for prior models anchor Fable to stale patterns, let it use its own judgment first. 4. Move from providing tasks to providing objectives. Describe what done looks like and how to verify it, then let Fable find the path (/loop and /goal are built for this)

1hViews 803Likes 4Bookmarks 1
Rich Kuo@richkuo7

@alexalbert__ ohhh nice ty, i was at 99% and was like, what am i gonna do for the next 4 hours and 30 minutes lol

2hViews 889Likes 7
13yr old with a credit card@13yroldwithcc

@alexalbert__ Can you give examples of how to use Fable? Every time it reads `unsafe {}` in my Rust code it defaults to Opus 4.8. It won't even chat to me it immediately disengages

1hViews 118Likes 3
The Opinionated Human@TheOpinionatedH

@alexalbert__ Please keep it in subscriptions after June 22nd. It's cheaper than Opus 4...

1hViews 94Likes 2
Alex Albert@alexalbert__

@richkuo7 🙌

2hViews 747Likes 3
Prithvish ⋅ d4mr@earth_ish

I am beyond impressed by its capabilities (as is everyone else). but using it is proving impossible because of the guardrails. especially because of the kind of product I am using it to build, even the most benign prompts are getting rejected.

please fix the safeguards before the model is removed from subscription

1hViews 164
Kenny Lamoot@kennylamoot

@wxrrjxr @alexalbert__ Rarely get near it. Subagents plus /clear, and I ask the model to write the handoff prompt for the next fresh session.

1hViews 25Likes 1
Petio Lazarov@petiosz

@alexalbert__ Okay but what about the bug where it just treats everybody as a criminal and wont do basic every day tasks, because it thinks they are unethical??

2hViews 518Likes 2
Wxrrjxr@wxrrjxr

@kennylamoot @alexalbert__ How does ur cable even do anything useful to get to 400k 😂😂

Every damn engineering prompt is blocked

1hViews 34
calum dew@calumjdew

@alexalbert__ After running Fable for 16 minutes i maxed my current usage window and used 21% of my total monthly sub with zero results... wth

1hViews 27
Dr Argy Bargy ⛩️@DrArgyBargy

@alexalbert__ I feel like the rate limits might need some adjustment. This ran out pretty fast.

1hViews 178Likes 2
Kenny Lamoot@kennylamoot

@wxrrjxr @alexalbert__ Not yet. But my work is product code, refactors, agent orchestration, nothing security-adjacent.

1hViews 3Likes 1
Justin@JustinPerea

@alexalbert__ I don't want to remove my Claude md for my other models. Can you set a global Claude Md just for Fable?

2hViews 314Likes 1
Emil@jivishov

@alexalbert__ Could you please look into the safeguards as well? Can't test it if it takes simple cases too seriously.

1hViews 259Likes 1
Wxrrjxr@wxrrjxr

@kennylamoot @alexalbert__ A better question if you have tried it for a bit now would be: did you get hit with any of the safety blockers yet?

1hViews 7
Kenny Lamoot@kennylamoot

@alexalbert__ Any tips on auto-compact window for Fable 5?

Kept mine at 400k because Opus degraded as context filled up.

But if Fable 5 hallucinates less at high fill, maybe that's too aggressive now. 🙏

2hViews 579
● goodtek@goodtek_xyz

@alexalbert__ Usage reset is clutch. Tip 1 hit home — I handed off a full MCP workflow slice instead of micro-prompts. High effort default means way less babysitting mid-run.

1hViews 508
Eric Buess@EricBuess

@alexalbert__ Thank you!

2hViews 140Likes 1
Anathema@zxc_2022

@alexalbert__ Bro, this is total BS.. I just subscribed to $200 pro plan and got this flag for absolutely no reason and now it won't continue working. how do I report this?

FML false flag on first prompt is actually insane, I'm not doing anything even close to cybersecurity or biology wtf?!

1hViews 101Likes 1
Load more posts