Claude Code hands paid users a limit reset through Oct 22

If you have ever hit Claude Code's 5-hour wall halfway through a refactor, you now have a spare. According to the announcement posted on Twitter, Pro, Max and Team users get a limit reset they can use whenever they choose, along with a cheaper default model, Opus 5.5.

At a glance

• Opus 5.5 becomes the default model on paid plans in Claude Code, and 5-hour session limits rise 20% starting today, alongside a single reset found in Settings → Usage.
• On the Claude Platform, Opus 5.5 costs $4/$20 per million tokens, down from $5/$25 for Opus 5, while cache reads fall 60%, from $0.50 to $0.20.
• The catch: Opus 5.5 carries Fable 5.1-class safeguards on cyber, bio and frontier LLM development, so flagged requests fall back to another model, and some flags are incorrect.

If you have not been following, Claude Code meters usage in two windows, a 5-hour session limit and a weekly limit, and once one runs out you wait. The reference point for every saving in this announcement is Opus 5, which listed at $5/$25 per million tokens on the Claude Platform.

In Claude Code, three things change at once. Session limits in the 5-hour window go up 20% today. Opus 5.5 is priced below Opus 5, so according to Anthropic your 5-hour and weekly limits go 25% further on it. And Pro, Max and Team subscribers get a reset in Settings → Usage that can be applied any time until Oct 22.

Anthropic says Opus 5.5 performs at the level of Fable 5.1 while running about 30% faster and costing about 40% less per task than Opus 5. The company also says it writes better: clearer, with less jargon, and more faithful to the writing rules you give it, so long Claude Code sessions are easier to follow. Anthropic calls it its new daily driver for coding and agentic work.

The model is available now in Claude Code and on the Claude Platform. Switching takes one command: run /claude-api migrate inside Claude Code to move to claude-opus-5-5. If you need more speed, fast mode goes up to 2.5x and costs $8/$40 per million tokens, double the standard Opus 5.5 rate.

Anthropic describes safeguards as one way it prevents misuse as its models grow more powerful. Opus 5.5 is the first Opus with Fable 5.1-class safeguards on cyber, bio and frontier LLM development. Requests that get flagged fall back to another model, and the company says it is working to reduce incorrect flags.

Two settings matter for long agent runs. Thinking is always on for Opus 5.5, with medium as the default, and you can adjust the effort parameter. For long-running agents, compaction summarizes older turns while keeping recent ones intact, much like a colleague who keeps the last few emails of a thread in full and a one-paragraph summary of everything before them.

The announcement leaves gaps. It does not name the model that takes over flagged requests or say how often flags misfire, and the 25% stretch only holds while you stay on Opus 5.5. In our view, a fallback to an unnamed model is the detail heavy users of security tooling in Claude Code should watch, because a misfire may mean the rest of a task runs on something other than what they chose.

Before the October 22 cutoff

The reset has to be applied by Oct 22, so the practical decision is which bad week to spend it on. Anthropic gives no timeline for cutting incorrect safeguard flags and does not say whether the 20% session increase is permanent. It also has not said whether more resets will follow once this one expires.