Anthropic will keep Claude Fable 5 inside Max and Team Premium plans from July 20 at 50% of their usage limits, while Pro and Team Standard customers move to separately billed credits. The durable change is who gets bundled access, not a larger allowance for top-tier users.
Anthropic is replacing a string of temporary Fable 5 extensions with a lasting division between subscription tiers. The top plans gain certainty, but not a larger disclosed allowance; lower tiers lose Fable as a bundled benefit.
Beginning July 20, Fable 5 will be included in all Max and Team Premium plans at 50% of their limits, Anthropic said in its announcement. Pro and Team Standard customers will instead access the model through usage credits, with a one-time $100 credit from Anthropic.
| Plan | Fable 5 access from July 20 | What is bundled |
|---|---|---|
| Max | Included | Up to 50% of the plan's usage limits |
| Team Premium | Included | Up to 50% of the plan's usage limits |
| Pro | Usage credits | No recurring Fable allowance; one-time $100 credit |
| Team Standard | Usage credits | No recurring Fable allowance; one-time $100 credit |
The percentage is not a new halving of Max and Team Premium access. After Fable returned on July 1, paid-plan access was already limited to 50% of weekly usage, according to a chronology of the rollout. The July 20 policy makes that capped access standard for two tiers and removes the bundled benefit from the other two.
Nor does “50% of limits” tell customers how many tokens or tasks they receive. Anthropic's post gives no fixed allowance, and Fable shares a plan's usage pool with other Claude models, as an account of the plan mechanics explains. A separate report says a bonus-usage period also ends July 20, reducing regular limits by 33% before the Fable cap is applied. The company post does not define that comparison or confirm the calculation, so it should not be combined with the 50% figure into a precise effective cut.
For Pro and Team Standard customers, “continued access” does not mean continued inclusion. Anthropic's credit documentation says usage credits are charged separately from a subscription at standard API rates. For individual paid plans, users must enable credits and set up payment, can impose a monthly spending cap, and receive a warning before usage switches to the separate balance; the retained documentation does not describe the Team setup flow.
Anthropic lists Fable 5 at $10 per million input tokens and $50 per million output tokens in the model announcement. At those rates, $100 is enough for at most 10 million input-only tokens or 2 million output-only tokens. Those are boundaries, not workload forecasts: ordinary sessions use both input and output, and the retained sources do not provide a representative token mix.
The distinction also matters for product positioning. Fable is Anthropic's generally available Mythos-class model, but some requests in cybersecurity, biology, chemistry and suspected model distillation fall back to Opus 4.8. Anthropic said at launch that the fallback occurred in fewer than 5% of sessions on average. Customers are therefore paying for access to a guarded service, not an unconditional guarantee that every request runs on Fable.
Anthropic launched Fable 5 on June 9 with included access for Pro, Max, Team and seat-based Enterprise plans through June 22. It said demand would be difficult to predict and planned to move subscribers to usage credits afterward unless capacity allowed an extension; its stated aim was to restore Fable as a standard subscription feature when sufficient capacity arrived.
The schedule was interrupted when Fable became unavailable on June 12 amid US export controls. The controls were lifted June 30 and the model returned July 1, according to the published timeline. Included access was then due to end July 7 before Anthropic extended it to July 12 and again to July 19. A contemporaneous account records the two post-relaunch extensions.
In follow-up comments quoted by those accounts, Anthropic acknowledged that the shifting limits had frustrated customers and said it wanted to make plan benefits more certain. That claim-origin chain matters: the frustration and capacity explanation are Anthropic's characterization, not an independently measured cause of the limits.
The July policy arrives as customers have more substitutes. GPT-5.6 entered general availability on July 9 and Moonshot AI released Kimi K3 on July 16, according to the same account. The timing does not establish that competition caused Anthropic's decision.
It does sharpen the customer's price comparison. Artificial Analysis, which said it supported OpenAI with prerelease evaluation, scored GPT-5.6 Sol at maximum reasoning one point below Fable 5 on its Intelligence Index, at an estimated $1.04 per task—about one-third of Fable's measured cost in that test. The evaluation also found Fable ahead on its AA-Briefcase knowledge-work benchmark, with a 56% rubric score versus 42% for Sol and a higher analytical-quality rating.
Those results are workload-specific, not a universal ranking. They show that a lower measured cost can come with near-parity on one composite index while Fable retains a substantial lead on another test. For subscribers deciding whether to upgrade plans or buy credits, the relevant comparison is their own task mix—not a claim that either model is categorically cheaper or better.
Anthropic's capacity explanation sits beside an unusually large infrastructure and financing package. In April, the company said it had committed more than $100 billion over 10 years to AWS technologies for up to 5 gigawatts of new capacity. It also said it already used more than one million Trainium2 chips, while Amazon was investing $5 billion immediately, with up to another $20 billion in the future on top of $8 billion previously invested. All of those figures come from Anthropic's joint-compute announcement and are company commitments or claims, not independently verified delivery totals.
Co-founder and CEO Dario Amodei said the infrastructure was needed to keep pace with demand. Anthropic projected “meaningful compute” within three months of the April 20 announcement and nearly 1 gigawatt in total by the end of 2026. The first window reaches roughly July 20—the same day the new Fable rules begin—but the announcement does not allocate that capacity to consumer subscriptions or promise that it will raise their limits.
That missing bridge is the central constraint on the story. A multiyear capital commitment and future gigawatts do not reveal how much Fable capacity is available now, how Anthropic distributes it among API, enterprise and consumer demand, or whether the 50% cap reflects infrastructure, pricing strategy, or both.
The immediate test is visible in customer accounts: the actual Max and Team Premium allowance after the bonus period ends, the point at which Fable usage stops, and whether the usage dashboard separates that cap clearly from other Claude consumption. Those observations can resolve the ambiguous 33% claim and show what “50% of limits” means in practice.
The longer-term test is whether new compute changes the entitlement. Anthropic has not announced a threshold for raising the top-tier cap, restoring bundled access to Pro and Team Standard, or lowering Fable's metered price. Until it provides one—or customers see materially larger allowances—the July 20 policy should be read as a durable allocation rule: it makes access predictable for the highest tiers while leaving the underlying scarcity and its economics undisclosed.
Christian Faith Madison’s estate alleges GPT-4o sustained a months-long narrative of prophecy, sacrifice and resurrection before her death. Public safety disclosures and controlled research make that failure mode plausible, but neither establishes which model behavior she encountered or whether it caused her death.
Shanghai AI Laboratory has released Intern-S2-Preview-397B as downloadable weights and a hosted API, but its own 35B alternative offers the same context window and a far smaller disclosed scale while the larger model still lacks auditable comparative scores, pricing and physical-validation results.
Mixfont's free Decoy Font gives each glyph a sharp decoy and a blurred intended letter, creating a hurdle for pixel-based readers while leaving selectable text, known-image techniques and accessibility costs outside its protection.
Zhipu reportedly reached $1 billion in annual recurring revenue in July, roughly four times a March estimate, but the unconfirmed run rate is not annual sales and still sits far ahead of recognized cloud revenue while margins remain thin.
Cmpunlocker says patched Nvidia open kernel modules can expose 64GB of HBM on an 8GB CMP 170HX or 40GB on a 10GB card while removing an SM throttle. The exploit mechanism is documented, but the retained evidence does not independently validate the tool across full-memory workloads, error rates, power use or A100-class performance.
An unidentified industry source says at least one Nvidia board partner has physical RTX 50 Super cards but cannot sell them, with 3GB GDDR7 quoted at $60 to $70 per chip. The reported memory bill explains a plausible pricing problem, but neither Nvidia nor a second independent source has confirmed the products, the hold or the component terms.
Anthropic’s 20-year lease gives TeraWulf a customer for 401 megawatts of future AI infrastructure, but rent starts only after delivery and the proposed utility deal passes power and grid costs to TeraWulf, leaving financing and project margin unresolved.
OpenAI has put conversations and Projects back in its redesigned ChatGPT desktop app and enabled cloud Work threads to move across devices, correcting the launch's biggest usability failures without merging local Work or Codex histories.
Twenty-nine countries signed an agreement creating WAICO as an independent intergovernmental organization, while China paired the launch with capacity-building offers that are not yet confirmed as WAICO programs.
Meta reportedly plans to put departing AWS compute executive Dave Brown to work on its data-center buildout, adding hyperscale operating experience while leaving any customer-facing cloud business conditional and undefined.
Moonshot AI has made Kimi K3 available through its apps and API, pairing a 2.8-trillion-parameter architecture with early frontier-level results, but the model's open-weight claim cannot be tested until its weights and technical report arrive.
CIA Director John Ratcliffe said US intelligence is consistent with an estimate that Russian recruits last 20 to 30 minutes on Ukraine’s battlefield, but the public trail leads to an unsourced claim about assault troops and does not establish a representative average.
China has paired a five-year AI training offer for developing countries with cooperation centers, a weather-warning rollout and a new 29-country organization. The package gives Beijing a platform for influence, but no budget, selection rules or delivery timetable has been published.
The FastFlowLM team has joined AMD after building a Ryzen AI NPU inference flow on technology AMD already developed and distributed. The move could reduce model-support delays, but its financial significance, transaction perimeter and performance benefit remain undisclosed.
Federal intervention changed the launches of Anthropic's Claude Fable 5, Claude Mythos 5 and OpenAI's GPT-5.6, but through different legal and voluntary channels. The result is real government leverage over early customers without a common security threshold or durable review process.
A proposal developed with Treasury Secretary Scott Bessent would put an independent AI regulator under SEC oversight, but President Trump had not reviewed it and its tests, funding and enforcement authority remained unresolved.
San Francisco has demanded that Apple and Google cut off 13 apps capable of producing nonconsensual sexual deepfakes. The threatened case could test California’s new liability rules, but the confidential app list and the law’s “primary purpose” definition leave a central question unresolved.
HKT plans a 3.2 Tbps data-centre route from Lok Ma Chau to Tseung Kwan O by the end of 2026, but has not disclosed customer bandwidth, price, project cost or end-to-end latency, while Sandy Ridge's developer expects to start within 42 months of its land award.
Apple briefly topped Nvidia during Friday trading before Nvidia closed about $6 billion ahead. The filings show why that ranking is a weak AI verdict: Apple’s operating momentum predates its new Siri, while Nvidia’s rapid growth comes with concentrated customers and large forward commitments.
Anthropic has proposed paying Meta as much as $10 billion over two years for AI computing capacity, but early exit rights, undisclosed capacity and Meta's own reliance on outside infrastructure leave the economics—and the case for a durable Meta cloud business—unresolved.