After 24 hours with Fable 5 back in the rotation, my read is more positive than the loudest reaction online. I understand the frustration, though.
If someone did not use Fable the last time around, and did not see how much agent work it can fan out under a serious request, the new limits can feel like a trap. It is not a normal chat model when you ask it to do big codebase work. It behaves more like a heavy production lane. That is powerful, but it means the plan tier and session shape are not side details. They are the product.
The usable plan reality
My practical take is blunt: Fable is most usable from the 20x tier. On a lower plan, especially a $100/month plan, a single Fable 5 Ultra Code run can chew through a five-hour window quickly if the project is large and the ask is not shaped well.
That does not mean the model is bad. It means the operator has to know what they are buying. A large Fable session is not where I would open a vague prompt and hope the model finds the work. I want the target named, the stopping point clear, and the surrounding chats arranged so one interrupted session does not waste the whole run.
I set my expectations that way before firing it up. When I started a serious Fable 5 session, it was not a directionless ask. It had a lane.
A Max 20x usage panel made the point concrete, but the real story was not one casual 21-hour burn. After a 2 p.m. reset, I used 100% of every five-hour window I could, including a midnight run and a 5:20 a.m. alarm to start the next window so it would reopen at 10:20. That left about 40 minutes of a fresh window before the 11 a.m. reset.
In practice, that was four full five-hour windows plus 34% of a fifth. That is what produced 83% weekly all-model usage and 100% Fable weekly usage. The 20x tier is where Fable feels usable, and it can still be maxed if the heavy lane stays hot. The lesson is not panic. It is calendar-aware planning.
The stack that felt efficient
The pattern that felt strong was not "run everything everywhere." It was a deliberate stack:
- one Fable Ultra Code lane
- one Fable High lane
- one Sonnet 5 Ultra Code lane
- two Sonnet High lanes
- one Opus Ultra Code lane
- two to four Opus High lanes
That kind of setup can keep five to seven chats moving at once, but only if the roles are distinct. Fable gets the largest bounded work. Sonnet sharpens, reviews, and closes. Opus handles deeper reasoning or parallel interpretation where it fits.
The risk is session collision. If too many heavy chats are active at the same time, sessions can stop midstream. That is where the waste happens: lost work, stranded context, and tokens spent on recovery instead of progress.
The safer pattern is to leave yourself a backup lane. It may look inefficient to keep capacity idle, but it is cheaper than getting stopped in the middle of a useful run.
The learning curve is real
There is a learning curve in how to ask for work now. You have to layer the asks.
Start with the goal. Let one model shape the work. Move the heavy lift into the best lane. Bring a second model in for review or closure. Before ending the chat, ask what remains, what should be implemented before stopping, and what a good handoff looks like.
That sounds simple, but it is a different habit from casual prompting. The people getting the most out of these tools are not just buying more tokens. They are learning how to operate sessions.
Why I am still satisfied
Overall, I am very satisfied with the Fable 5 release. I just wish the current 50% weekly-limit treatment would stay. If the terms stayed like this, the model would feel much easier to plan around.
The first day already produced big results for the digital landscape I am building. My plan for this launch week is to use the restored Fable window like a concentrated build sprint: roughly six weeks of normal limits compressed into one careful week of targeted work.
That is why my conclusion is not "the backlash is wrong." It is more specific: the backlash is understandable if someone expected a casual, always-on heavy model. My experience is that Fable 5 is excellent when treated as a scarce, high-leverage work lane.
The difference is preparation.
Related: July 1: Fable 5 is back, and session shape matters.