Anthropic's announcements page records three model releases in the Claude 5.5 line, issued within a few weeks of each other in late September and early October 2026.
What was released
Claude Opus 5.5 (Sep 22, 2026). Anthropic says it leads in agentic coding and knowledge work, and that it costs 40% less to run than Opus 5 on typical workloads.
Claude Sonnet 5.5 (Sep 28, 2026). Anthropic calls it a clear upgrade over Claude Sonnet 5. It is said to run more than 30% faster and to cost up to 30% less for most work.
Claude Haiku 5.5 (Oct 7, 2026). Anthropic describes it as its fastest, most capable small model, built for high-volume work such as summarization, subagents and browser use.
What the source does not say
The page carries only short summaries. It gives no per-token prices, context limits, benchmark scores or availability details for any of the three models, and it does not say which platforms or plans include them. The cost and speed comparisons are Anthropic's own claims, not independent measurements. Check the individual announcements before changing any production setup.
Why it matters for operations
The common thread is cost per unit of work. Two of the three releases come with a stated reduction against the previous generation, and the third targets the repetitive, high-volume steps where cost accumulates fastest. For a team running automated ticket handling, lead enrichment or document summarization, the practical step is a side-by-side test of the new and old models on a sample of real tasks, measuring quality and cost before switching.
The same page lists other Anthropic news from the period, including a Singapore office, funding and training programs. None of those changes a product a business uses, so they are not covered here.