Claude Code now ships with the Claude Code Opus 5.5 default model switch built in: update to version 2.1.280, released on 22 September 2026, and Claude Opus 5.5 becomes the default Opus model in every session, with a 1M-token context window, pricing of 4 dollars per million input tokens and 20 dollars per million output tokens, and cache reads at 0.20 dollars per million tokens. You do not need to change a single setting — the official changelog entry reads: "Added Claude Opus 5.5 (claude-opus-5-5), now the default Opus model — 1M context, $4/$20 per Mtok with $0.20/Mtok cache reads."
📺 Watch: Claude Opus 5.5 AI Just Changed Everything
🔥 Get the Agent OS as a free bonus: AI Profit Boardroom members get the full Agent OS zip, prompt libraries, daily tutorials and weekly live coaching calls. → Get inside · Want AI SEO help 1-on-1? Book a free SEO strategy session →
That one changelog line is the biggest practical change to Claude Code since Claude Opus 5 took over as the default Opus model in July 2026. If you run agents, content systems or coding workflows on Claude Code, the model doing the heavy lifting just got cheaper and fresher at the same time — and this page walks through exactly what changed, what it costs and what to do about it, with every figure taken from the Anthropic changelog and official model documentation fetched on 23 September 2026.
What The Claude Code Opus 5.5 Default Model Update Actually Changes
Three things change the moment your Claude Code installation moves to v2.1.280. First, the default Opus model becomes claude-opus-5-5 — so Opus sessions, subagents set to Opus and Opus-powered background work all pick up the new model without any configuration. Second, the pricing baseline drops: Anthropic's model documentation lists Claude Opus 5.5 at 4 dollars per million input tokens and 20 dollars per million output tokens, against 5 dollars and 25 dollars for Claude Opus 5, which is now marked as a legacy model. Third, cache economics improve — cache reads on Opus 5.5 cost 0.20 dollars per million tokens, compared with 0.50 dollars on Opus 5, which matters enormously for long agent sessions that re-read the same project context turn after turn.
The context window stays at 1M tokens and maximum output stays at 128K tokens, so nothing about your existing long-context workflows needs restructuring. Anthropic's models overview describes Claude Opus 5.5 as built "for long-running agentic coding and knowledge work", and lists a reliable knowledge cutoff of June 2026 — a month fresher than the May 2026 cutoff on Claude Opus 5. If you already run local sessions, the Claude Code local setup guide covers where the harness fits on your machine, and nothing in that setup changes with this release.
Claude Opus 5.5 Pricing In Claude Code: The Numbers That Matter
Per Anthropic's official model documentation, the comparison between the new default and the model it replaces looks like this: Claude Opus 5.5 costs 4 dollars per million input tokens and 20 dollars per million output tokens with 0.20-dollar cache reads, while Claude Opus 5 costs 5 dollars in and 25 dollars out with 0.50-dollar cache reads. That is a 20 percent cut on both input and output, and a 60 percent cut on cache reads, for the model Anthropic now recommends as the starting point for most workloads.
For anyone running high-volume automation through Claude Code — AI SEO pipelines, research agents, bulk content systems — the cache-read cut is the sleeper detail. Long-running agent sessions spend a large share of their token bill re-reading cached project context, and that line item just fell by more than half. The Claude Code AI SEO walkthrough shows the kind of pipeline where those cached-context savings compound across hundreds of runs. And if budget is the blocker, the guide to using Claude Code free covers the plans and trials that let you test all of this before spending anything.
If you want working agent systems built on updates like this — the day they drop, not months later — check out the AI Profit Boardroom → get the full Opus 5.5 playbook inside. Want a 1-on-1 plan for your own AI SEO setup first? Book a free SEO strategy session and map it out.
📺 Watch: Claude Just Dropped Game Changer Updates!
Everything Else In Claude Code v2.1.280 And The Releases Around It
The Opus 5.5 default is the headline, but the changelog around v2.1.280 carries several changes worth knowing. In v2.1.280 itself, fast mode — Claude Code's lower-latency way of running Opus — became available in Claude Code Remote sessions on cloud and self-hosted runners where organisations permit it, so cloud sessions can now get the same speed treatment as local ones. The same release added mouse-wheel support to the skills list and fixed a batch of keyboard-navigation issues across dialogs and settings panels.
Two earlier September releases set up the context. Version 2.1.277 added AGENTS.md support as an alternative to CLAUDE.md in projects, which matters if you standardise agent instructions across teams or repositories. Version 2.1.278 changed auto mode for Claude API and Enterprise users — and on Bedrock, Vertex, Foundry and gateways — to default to the server-side classifier that handles permission decisions. None of these require action, but together they show where Claude Code is heading: faster remote execution, more standardised agent configuration and less permission friction. The learn Claude Code guide is the right starting point if these terms are new to you.
How To Get The Claude Opus 5.5 Default In Your Claude Code Setup
Update Claude Code to version 2.1.280 or later through whichever channel you installed it with, and the new default applies on your next Opus session — there is no flag to set and no model ID to paste. If you have previously pinned a specific model in your settings or environment configuration, that pin still wins, so check your own configuration if you want the new default to take effect. Sessions already running keep their model until they end.
One honest caveat: a new default model is not automatically a better result on your specific workload. Anthropic's documentation states Opus 5.5's default effort level is medium, where Claude Opus 5 defaulted to high, so if you have effort-sensitive workloads it is worth testing both settings rather than assuming parity. For how the current crop of models compares in hands-on tests across real agent workloads, the Goldie Bench write-up covers how these brains stack up — that is the resource to check before you migrate anything mission-critical.
What The New Default Means For Agents And AI SEO Workflows
The practical read for anyone building money-making systems on Claude Code: your Opus-tier work just got roughly 20 percent cheaper per token while gaining a fresher June 2026 knowledge cutoff. Cheaper Opus changes the calculus on which tasks deserve the big model. Work you previously routed to smaller models purely on cost — long research passes, full-site content audits, multi-step agent chains — becomes viable on Opus 5.5 at the new pricing, especially with cache reads at 0.20 dollars per million tokens.
This also lands well for structured agent setups. If you run the Agent OS approach — persistent instructions, skills and memory wrapped around Claude Code — the default-model change slots straight in, because Agent OS workflows inherit whatever model the harness runs. The same applies to combined setups like the agentic OS on Claude Code configuration, where subagents doing delegated work now also default to the cheaper, fresher Opus. And Claude Code is not the only harness shipping updates this month — the Hermes Agent 0.21.4 update landed on 21 September with its own stack of changes, so the whole agent tooling space is moving at once.
Claude Code Opus 5.5: Questions Worth Answering
Is Claude Opus 5 gone? No — Anthropic's documentation lists Claude Opus 5 as an active legacy model, released on 24 July 2026, with retirement not sooner than 24 July 2027, and it remains selectable by its model ID. Do you pay more for the 1M context? The 1M-token window is the standard context for Opus 5.5 in the documentation cited above; Anthropic's pricing page covers long-context and per-platform specifics, so check it if you routinely push very large prompts. Does the update change Sonnet or Haiku defaults? The v2.1.280 changelog entry covers the Opus default only.
Where does this leave earlier Opus content and workflows? Working, but worth revisiting. Guides written around Claude Opus 4.7 for AI SEO still describe workflows that function today, but the pricing tables and model IDs in them describe a different generation — the techniques carry over, the economics have improved. That page remains useful for the workflow logic; this page is the current state of the default model itself.
The bottom line: v2.1.280 is the rare update that requires nothing from you and still cuts your bill. Update, confirm your sessions show claude-opus-5-5, re-run your heaviest workflow and compare the cost line — the changelog and Anthropic's model documentation, both cited and dated above, say that is exactly what should happen.
If you want to turn cheaper Opus tokens into actual revenue — agents, AI SEO systems and daily builds shared as they ship — check out the AI Profit Boardroom → join the builders inside. Rather talk it through first? Book a free SEO strategy session and get a plan for your site.
