OpenAI said it would restore Codex and ChatGPT Work users to 100% of their weekly usage limit after citing 9 million active users, following an Anthropic reset and a similar exchange a week earlier. The extra capacity is immediate, but neither reset changes the products’ documented standard allowances or resolves how the companies will price sustained agent workloads.
OpenAI and Anthropic have now announced broad usage resets twice in a week to give agent customers more capacity. The resets are valuable to users who have reached a limit, but they do not by themselves establish a new standard allowance or a lower long-term cost for agent work.
Tibo Sottiaux said OpenAI was performing “another reset” for Codex and ChatGPT Work users. In his July 16 post, he wrote that the products had hit 9 million active users earlier that day and that users should have their “100% weekly usage limit” back within minutes.
The wording matters: the post announced an imminent restoration, rather than documenting that it had been completed. It also did not define “active,” specify a daily or weekly measurement period, or divide the 9 million between Codex and ChatGPT Work.
Anthropic had announced a reset of Claude’s five-hour and weekly limits just before Sottiaux’s post, while OpenAI announced a reset of its weekly limits, according to a report on the sequence. The same report said the two companies had each reset limits on July 9, making July 16 the second such exchange in a week.
That chronology supports the description of a competitive response. It does not show that either company permanently raised its normal entitlement.
OpenAI has published several adoption numbers, but they measure different populations and periods. Its April team-pricing announcement said more than 9 million paying business users relied on ChatGPT for work and that more than 2 million people used Codex each week. At the July 9 launch of ChatGPT Work, OpenAI said weekly Codex use had passed 5 million, including more than 1 million people using it outside software development, in the product announcement.
Those disclosures show growth in weekly Codex use between April and July. They do not establish what changed when Sottiaux cited 9 million active users: the April 9 million figure covered paying business users of ChatGPT, the July 9 figures covered weekly Codex users, and the July 16 post did not define its interval or scope.
The reset also reaches products with potentially heavy workloads. OpenAI describes ChatGPT Work as an agent that can operate across apps and files and stay with a complex project for hours. The company says usage varies with the work required and that more complex tasks can consume more of a plan’s included usage. Restoring the weekly meter can therefore rescue an interrupted project, but its value depends on the user’s plan and workload.
Users welcomed the immediate benefit. Dan Shipper, CEO of AI publication Every, called the reset contest “capitalism stays winning!” The analysis carrying his reaction described the added usage as arriving without a price change, while emphasizing that OpenAI’s reset was temporary rather than a permanent relaxation of its ceiling.
OpenAI’s own offers show why a reset and a pricing change should be separated. In the same April announcement, the company said Codex-only team seats had no rate limits and were billed by token consumption, while standard ChatGPT Business seats included Codex usage limits. It also reduced the annual-plan price of a ChatGPT Business seat from $25 to $20 and offered eligible workspaces limited promotional credits for new Codex-only users.
OpenAI then stopped offering new pay-as-you-go Codex seats to Business plans on June 24, while leaving existing seats unaffected. Separately, OpenAI says Enterprise and Edu administrators can set ChatGPT Work controls at workspace, group and individual levels and review requests for additional credits. These are distinct ways to allocate agent capacity: a capped entitlement inside a seat, metered token spending, and discretionary or promotional credits.
A reset temporarily changes the first mechanism. It does not reveal the normal quantity of work represented by 100%, lower the token price of metered use, or guarantee that another reset will occur.
Anthropic’s earlier actions complicate the idea that limits are only a promotional lever. In a May announcement, the company doubled Claude Code’s documented five-hour rate limits for Pro, Max, Team and seat-based Enterprise plans, removed a peak-hours reduction for Pro and Max accounts, and raised API rate limits for Claude Opus models.
Anthropic tied those changes to a SpaceX compute partnership and other capacity deals. It said the SpaceX agreement would provide access to more than 300 megawatts of capacity, or over 220,000 Nvidia GPUs, within a month. Those are company claims, and the source does not provide independent confirmation that all of the capacity arrived on schedule.
The distinction is useful. A published baseline increase links customer entitlements to a stated infrastructure change. A broad one-time reset supplies extra work immediately but says nothing about whether efficiency, available compute or the provider’s willingness to absorb cost has changed.
The next consequential evidence would be a documented change to the ordinary plans: higher stated limits, a longer promotion with fixed terms, broader access to metered seats, lower credit costs, or clearer disclosure of how agent usage is measured.
OpenAI also needs to define the 9 million figure before it can serve as a meaningful adoption benchmark. A measurement period and a split between Codex and ChatGPT Work would allow comparison with the company’s weekly Codex disclosures; without them, it is a promotional milestone attached to a useful but temporary reset.
Until the companies publish new baseline terms, customers can treat the resets as bonus capacity—not as evidence that the recurring economics of long-running agents have changed.
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.
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.
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.
RoboTTT extends a robot policy’s visuomotor context to 8,000 timesteps and raises its average rubric score from 42% to 79%, but the evidence comes from three author-run assembly tasks and the longest task was completed in only two of ten trials.
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.
Nebius has arranged its first senior secured facility against an operating GPU deployment and one customer’s cash flows. The deal adds project-level debt to its expansion toolkit, but an unnamed customer and a larger rival financing limit what it proves about the rest of Nebius’s backlog.
Cursor says automatically running a repository-root git.exe on Windows does not meet its criteria for patching, while the researcher calls it an untrusted-search-path defect and separate research shows the same weakness across several competing AI coding tools.
Huawei publicly displayed a 16-cabinet Atlas 950 configuration rated at 1 EFLOPS in FP8, providing tangible evidence of its system-scale AI strategy while leaving price, power use and sustained workload performance undisclosed ahead of the full system's planned fourth-quarter release.
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.
UK testing places leading open-weight models four to seven months behind selected closed-model cyber results, yet longer attack chains, U.S. benchmarks and mixed cost comparisons show why that interval is a warning signal rather than a universal capability clock.
TSMC reached the top of its second-quarter revenue guidance and raised its 2026 outlook, but a one-off investment gain boosted profit growth while the company committed more capital to 2-nanometer production, advanced packaging and an undated Arizona expansion.
Moonshot AI's Kimi K3 added to a global technology selloff with near-frontier performance, but unreleased weights, mixed cost comparisons and a recommended 64-accelerator deployment leave its effect on chip demand unresolved.
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.
Two binding EU decisions require Google to give rival AI services comparable access to 11 Android features and offer eligible search competitors a restricted, anonymized dataset, but phased deadlines, certification, pricing and privacy safeguards leave the competitive effect unproven.
Databricks has signed a term sheet for a Coatue-led financing at a $188 billion valuation, while unidentified sources put the round at $3 billion. The proposed capital would deepen its push into AI governance, data agents and operational databases, but the transaction remains open and the company supplied no new operating figures.
Netflix says generative AI workflows were used on roughly 300 titles in 2026, but its only quantified example tied to that disclosure covers 17 minutes and does not establish how much finished material, spending or labor changed across the slate.
Gemini 3.5 Pro missed its expected June rollout. An anonymous-source account says a late-June data change fell short of Google's coding goals, but Google has confirmed only partner testing—not the reported cause, a new date, or public results and pricing.
Google is bringing Gemini Omni editing and a reusable face-and-voice avatar to Vids, but the sharper distinction is account-level identity across Vids and Gemini rather than a new category of AI video; Vids already offered Veo generation and customizable avatars, while specialist rivals already sell digital presenters.
Intel plans to adopt Gemini Enterprise across engineering, supply-chain and corporate work while extending chip-simulation capacity into Google Cloud, but the companies supplied no rollout schedule, contract value or performance baseline.
The Navy has approved an immediate department-wide framework for turning data into operational effects, but the public rollout leaves budgets, owners, deadlines, performance measures and assurance rules to a promised implementation roadmap.