AI News
All stories
Reported

Theo says Claude plans add a separate Fable 5.2 weekly limit and account-age resets

Theo says the $200 Claude plan now has a general weekly limit plus a separate 7-day Fable limit, and that model resets restore half the weekly allowance.

Products · · 3 min read

Theo says Anthropic's $200 Claude plan now splits usage into two limits: a general weekly limit and a separate seven-day limit for Fable 5.2. Under that arrangement, the Fable limit can drop to zero while half of the weekly allowance is still available. These mechanics come from Theo's own test across several of his accounts, and he presents his figures as his own calculations.

The two limits are linked by a conversion rate. Theo says every two points of Fable usage count as one point against the general weekly limit, so the Fable limit is an extra cap layered on top of the weekly allowance, not a separate pool of tokens. Many users assume that an empty general limit and a full Fable limit means they can keep using Fable, Theo says, but they cannot. He used to leave accounts in exactly that state, with Fable exhausted and half the week unused, because Opus 5 was weak while Fable 5 and 5.1 were strong. Now that Opus 5.5 can use the leftover half, he says he burns his accounts down far more often.

Theo says Anthropic granted Fable resets according to account age: one for accounts under six months old, two for accounts six to twelve months old, and three for accounts over twelve months old. One of his accounts received three, and the others received one or two. In his test, he had 22% of Fable and 11% of the overall limit left, and he used a reset. Afterward the weekly figure showed 51% used and the Fable figure showed 0%, so the reset restored roughly half of the weekly allowance and cleared the Fable limit. The reset also expires in seven hours. Unlike Codex resets, which Theo says reset the weekly timer to seven days, the Claude reset left the timer alone, which he finds easier to plan around. The video does not say whether these account-age resets will continue after this release.

On the economics, Theo estimates that a month of maximum Opus 5.5 usage on the $200 plan is worth about $14,000 at API prices, which he puts at 73 times the plan's price. If a user spends the whole Fable allowance, he says the multiplier falls to about 15x, an 80% decrease. In his split, about $3,000 of Fable usage comes first, followed by about seven grand of Opus 5.5 for the other half of the week. Fable 5.2 is cheaper and more efficient than its predecessor, he notes, so each dollar buys more output tokens. He says Fable gives about 20% fewer turns than going all in on Opus, but because Fable is more likely to finish a thread in a single turn, the mix gave him more threads overall. These are his estimates, and the video does not break down how he calculated the API-equivalent value.

Anthropic's listed API price for Fable 5.2 is $8 per million input tokens and $40 per million output tokens. An optional fast mode costs twice as much for a claimed 2.5 times the speed. Theo reports measured output speeds of 106 tokens per second for Fable 5.2 and 112 for Opus, compared with 81 to 84 for Fable 5.1. From the speed and price changes, he concludes that the model is probably smaller than Fable 5.1 or easier to host. He presents this as inference rather than something Anthropic stated.

Quality is harder to judge. Anthropic's benchmarks show large gains, including a state-of-the-art score on Artificial Analysis and a jump on agentic science research. Theo's own evaluations tell a different story. In his codebase analysis, Fable 5.2 made four wrong claims out of 61 about his TypeScript-to-Rust project, while a model he calls Soul made no false claims out of 38 in 29 minutes for $4.19. On his T3 Code analysis, Fable 5.2 made six false claims out of 83, and it took roughly 120 minutes to review eight pull requests, which he says undercuts much of its value for him. He says the model is more willing than others to let minor nitpicks slide, which he counts as both a strength and a weakness.

What changed

The $200 Claude plan now splits usage into a general weekly limit and a separate seven-day limit for Fable 5.2, where every two points of Fable usage count as one point against the weekly limit.

Type: Pricing and limits · Applies to: $200 Claude plan, Fable 5.2

Why it matters

Heavy users planning around Fable need to track two limits, because Fable usage also draws down the general weekly allowance. Teams that depend on Claude subscriptions should plan capacity around both.

What you can do

Check the usage meters on each Claude seat before assigning heavy Fable work.

Original source

We have no link to the original source for this story yet.

Where we saw it

1 source

Theo - t3.gg

AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works