Limit changes

Support page updated to include Haiku 5.5 references

Detected October 8, 2026, 05:15 UTC · Context window on paid plans

Anthropic updated its documentation for paid context limits to include references to Haiku 5.5, while removing an article link regarding automatic model changes during Sonnet 5.5 conversations.

This support document outlines the context window allowances provided across Claude's paid subscription tiers. In the latest revision, the reference material incorporates Haiku 5.5 in multiple places and includes a navigation link regarding how artificial intelligence outputs are labeled. Concurrently, the page drops a previous reference explaining why an active dialogue might be switched to an alternate model while using Sonnet 5.5.

In practice, premium account holders reviewing the document will now see Haiku 5.5 included in the listing of supported models for paid context capacities. Subscribers looking for troubleshooting details about mid-conversation model changes will no longer find that topic connected directly to this guide, as navigation has shifted toward documentation covering synthetic content attribution.

From the provided text alone, several practical boundaries remain unclear. The diff does not state the exact token maximum allotted to Haiku 5.5, nor does it confirm whether its context allowance matches or diverges from other options on the platform. Additionally, it cannot be determined whether policies governing automated model downgrades under high demand were modified or simply removed from this specific reference page.

We compare this page once a day and record every difference. The text above is written by Claude Network from that comparison; none of it is copied from the source.

Source: Context window on paid plans