Good news: Claude has announced a “reset of weekly usage,” set to suppress competitors on the day their new models are all released.

Anthropic’s official account @ClaudeDevs made an unannounced announcement at midnight today (10th), resetting the 5-hour and weekly usage caps for all users. The timing happened to line up exactly as OpenAI, SpaceXAI, and Meta released new models on the same day.
(Previously: Musk’s SpaceXAI officially launches its most powerful model “Grok 4.5”! Teaming up with Cursor to aggressively target AI coding agents)
(Additional background: Anthropic’s valuation surges 550% to $1.2 trillion, surpassing OpenAI! A frenzy in the secondary market to sell homes—everyone wants a piece)

On July 9, OpenAI, SpaceXAI, and Meta all unveiled new models one after another, while Google also posted the release schedule for Gemini 3.5 Pro on the wall—another lively round of the model arms race kicked off. Just as the community was analyzing which model is better / cheaper, Anthropic’s official account @ClaudeDevs at midnight on X made an unannounced announcement: **the 5-hour and weekly usage limits for all users have been reset—**the nearly depleted quotas were instantly replenished.

We've reset 5-hour and weekly rate limits for all users.

— ClaudeDevs (@ClaudeDevs) July 9, 2026

Rivals all roll out new products the same day

Anthropic chose to reset right at this moment—its timing is especially delicate.

That same day, OpenAI publicly released the GPT-5.6 family, bringing out three models at once: Sol, a flagship focused on frontier reasoning and long-horizon agentic tasks; Terra, a balanced option with its price cut to half of the prior GPT-5.5; and Luna, the fastest and cheapest.

Musk’s SpaceXAI also rolled out Grok 4.5 in parallel, targeting AI coding agents and integrating Cursor. It claims feature parity with Claude Opus, but with faster speed and a massive price cut.

Meta also released the agent model Muse Spark 1.1 with a million-token context, while Google previewed Gemini 3.5 Pro to arrive on July 17, touting a 2 million-token context and a Deep Think reasoning layer.

How the community sees it

Although the official account said nothing about the reason for the reset, it’s not hard to infer the intent behind resetting usage right when competitors are launching new models: so developers can keep using Opus 4.8, Fable 5, and several other flagship models without friction, preventing anyone from hitting the quota floor and temporarily deciding to try out GPT-5.6 or Grok 4.5.