Claude Opus 4.8 went live on May 28, 2026. But unlike a phone OS update, it does not just appear everywhere at once. Depending on which surface you use, you either get it automatically, have to update a CLI first, or have to change a model ID string by hand.

The single most common confusion: you open Claude Code, you’re still on 4.7, and Opus 4.8 isn’t even in the model picker. That is almost always a stale CLI version, not an account problem — and the fix is one command.

This is the complete switch guide for every surface, with the exact commands and the one model-ID gotcha that silently breaks API calls.

Key facts:

• Anthropic API model ID: claude-opus-4-8 — dashes, not a dot. claude-opus-4.8 is invalid and returns a 404 / not_found_error.
• Claude Code requires v2.1.154 or later to see Opus 4.8. Run claude update first. (Source: Claude Code model configuration docs)
• claude.ai and the desktop app: Opus 4.8 is in the model picker now; no update needed.
• The effort parameter defaults to high on every surface, including the API and Claude Code. (Source: Anthropic models overview)
• If 4.8 is missing from your Claude Code picker, the cause is a stale CLI version 95% of the time.

The 30-second answer

Where you use Claude	How to switch to Opus 4.8
Claude Code	claude update, then /model and pick Opus 4.8
Claude Code (one session)	claude --model claude-opus-4-8
claude.ai / desktop app	Click the model name in the prompt box → Opus 4.8
Anthropic API	Set model to claude-opus-4-8 (dashes, not a dot)
Cursor / Cline / other IDE tools	Update the extension, then select Opus 4.8 in its model menu
GitHub Copilot	Already GA — pick Claude Opus 4.8 in the Copilot model dropdown

If you only do one thing: in Claude Code, run claude update and then /model. That fixes the “4.8 isn’t in my list” problem for almost everyone.

For what actually changed in 4.8 and whether it’s worth switching, see our companion piece: Claude Opus 4.8 — the 4 changes builders should actually care about.

Claude Code: the full walkthrough

This is where most people get stuck, so it gets the most detail.

Step 1 — Check your version

claude --version

You need v2.1.154 or later for Opus 4.8 to be available. If you see anything lower (or 2.1.15x where the last digits are below 154), that’s your problem — 4.8 literally isn’t in the build you’re running.

Step 2 — Update the CLI

claude update

This pulls the latest Claude Code. Re-run claude --version afterward to confirm you’re now on 2.1.154+. (At time of writing, the current line is 2.1.156.)

This single step is the fix for the most common complaint: “I opened Claude Code, it was still on 4.7, and switching models didn’t even list 4.8.” The model picker only shows what your installed CLI version knows about. An old CLI has no idea 4.8 exists.

Step 3 — Open the model picker

Inside a Claude Code session, type:

/model

You’ll get an interactive picker that looks like this:

 Select model

❯ 1. Opus 4.8 (1M context)   ✔ recommended
  2. Opus 4.7
  3. Sonnet 4.6
  4. Haiku 4.5
  5. Default (recommended)

  Esc to cancel

Use arrow keys, hit Enter on Opus 4.8. As of v2.1.153, selecting a model in /model saves it as the default for all new sessions — it writes the model field into your user settings, so you don’t have to re-pick every time. (Source: Claude Code model configuration)

/model is also the most reliable way to confirm what you’re on, because it shows whatever the server is currently serving your account — not a cached guess.

Step 4 — Or switch for a single session

If you just want to try 4.8 for one run without changing your default:

claude --model claude-opus-4-8

Note the dashes. The CLI also accepts the short alias opus for “the latest Opus,” but pinning the explicit claude-opus-4-8 is safer if you want to stay on this exact version after 4.9 ships.

Step 5 — Or pin it in settings

To set it permanently without the picker, add the model field to your Claude Code user settings (~/.claude/settings.json):

{
  "model": "claude-opus-4-8"
}

This is the same field /model writes for you. Editing it by hand is useful for dotfiles, shared team configs, or CI environments where you can’t run an interactive picker.

Step 6 — Verify

Run /status inside a session, or /model again and look for the checkmark next to Opus 4.8. If it’s there, you’re done.

A note on Dynamic Workflows: the new parallel-subagent feature that ships alongside 4.8 is gated to Claude Code for Enterprise, Team, and Max plans only. Switching the model to 4.8 does not unlock it on lower tiers — that’s a plan gate, not a model gate.

claude.ai and the desktop app

No update needed here. Anthropic rolls the model out server-side.

1. Open claude.ai (or the Mac/Windows desktop app).
2. In the prompt box, click the model name at the bottom-right (it’ll say “Opus 4.8”, “Sonnet 4.6”, or whatever you last used).
3. Pick Opus 4.8 from the dropdown.

The claude.ai picker. Opus 4.8 sits at the top (“Most capable for ambitious work”). The “Effort: High” row and “More models” submenu are new with the 4.8 generation.

Two things to notice in that picker:

• Effort: High. This is the effort slider. Opus 4.8 defaults to high. You can push it to extra (called “xhigh” in Claude Code) or max for hard one-shot problems, but high is the right default for almost everything — the cost-quality curve flattens fast above it.
• More models. Older versions (Opus 4.7, Sonnet 4.5, etc.) live behind this submenu if you need to pin to one for a reproducibility reason.

The Anthropic API: mind the dash

For direct API use, switching is a one-line change — set the model field to:

claude-opus-4-8

The gotcha that bites everyone: it’s claude-opus-4-8 with dashes, not claude-opus-4.8 with a dot. A dot returns a 404 not_found_error and people waste 20 minutes assuming their API key is broken. Starting with the 4.6 generation, Anthropic switched to a dateless, all-dashes format (claude-opus-4-6, claude-opus-4-7, claude-opus-4-8) — each one is a pinned snapshot, not an evergreen pointer. (Source: Anthropic models overview)

Minimal curl to confirm it works:

curl -s https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-8",
    "max_tokens": 100,
    "messages": [{"role": "user", "content": "Reply with the model you are running."}]
  }'

In the official SDKs it’s the same string:

# Python
import anthropic
client = anthropic.Anthropic()
msg = client.messages.create(
    model="claude-opus-4-8",
    max_tokens=100,
    messages=[{"role": "user", "content": "hello"}],
)

// TypeScript
import Anthropic from "@anthropic-ai/sdk";
const client = new Anthropic();
const msg = await client.messages.create({
  model: "claude-opus-4-8",
  max_tokens: 100,
  messages: [{ role: "user", content: "hello" }],
});

If you use prompt caching or mid-conversation system messages, update to the latest SDK — older SDK versions reject the new message shapes. (Details in the what-changed writeup.)

Cloud platforms: AWS Bedrock, Vertex AI, Microsoft Foundry

Opus 4.8 is available on AWS (via Amazon Bedrock), Google Vertex AI, and Microsoft Foundry. The exact model-ID string differs per platform, and the safest move is to read it from each platform’s model catalog rather than typing it from memory.

The naming follows the same dateless dash pattern as 4.7:

• Claude Platform on AWS uses the same IDs as the Claude API — i.e. claude-opus-4-8.
• Amazon Bedrock (Bedrock-style IDs) follows the anthropic.claude-opus-4-x pattern that 4.7 used (anthropic.claude-opus-4-7).
• Vertex AI follows the claude-opus-4-x pattern 4.7 used (claude-opus-4-7).

Confirm the exact current string in your platform’s model list before shipping — cloud model IDs occasionally carry version suffixes (-v1:0) that the first-party API does not. (Source: Anthropic models overview, Claude Opus 4.8 on AWS)

Note: on Microsoft Foundry, Opus 4.8 has a 200k-token context window, not the 1M you get on the first-party API.

Cursor, Cline, and other third-party IDE tools

These wrap the Anthropic API, so two things have to line up:

1. The tool’s model list has to include 4.8. That ships in an app/extension update — so update Cursor / Cline / your tool of choice first.
2. The underlying SDK has to be recent enough to accept any new message shapes (mid-conversation system entries, etc.). If the tool errors on those, it’s running an old SDK and you wait for its next release.

In practice: update the extension, then pick Claude Opus 4.8 from its model dropdown. If it’s not listed yet, the tool hasn’t shipped support — there’s nothing to change on your end except wait or, if the tool lets you type a raw model ID, enter claude-opus-4-8.

GitHub Copilot

Claude Opus 4.8 is generally available in GitHub Copilot as of the launch. (Source: GitHub Changelog) Open the Copilot model picker (in VS Code, the chat panel’s model dropdown) and select Claude Opus 4.8. If you don’t see it, update the GitHub Copilot extension.

Which effort level should you use?

Once you’re on 4.8, you get the effort control. Short version:

Effort	Use it for
high (default)	Almost everything — coding, agent loops, normal chat
extra (“xhigh” in Claude Code)	A single hard sub-step: an architecture decision, a tricky refactor, a planning turn
max	Hardest one-shot problems with no time pressure; research-grade reasoning

The clean pattern in Claude Code: bump to xhigh for the planning turn of a multi-step task, drop back to default for the execution turns. The plan benefits from more thinking; the execution usually doesn’t. Don’t run your whole session on max — you’ll pay for thinking you don’t need.

Troubleshooting

Symptom	Cause	Fix
Opus 4.8 not in Claude Code /model list	Stale CLI version	claude update, then /model
Still on 4.7 after opening Claude Code	Old default pinned, or old CLI	claude update; /model → Opus 4.8
API returns 404 not_found_error	Used claude-opus-4.8 (dot)	Use claude-opus-4-8 (dashes)
SDK errors on new message fields	Old SDK version	Upgrade anthropic (Python) or @anthropic-ai/sdk (TS)
Dynamic Workflows not available	Plan gate, not model gate	Requires Claude Code Enterprise / Team / Max
4.8 missing in Cursor / Cline	Tool hasn’t shipped support	Update the extension; wait if not listed
claude.ai shows old model	Cached UI	Reload the page; click the model name and reselect

Bottom line

For 90% of people the whole “how do I switch to 4.8” question is solved by one of two things: in Claude Code, run claude update and then /model; everywhere else, just pick Opus 4.8 in the model menu. The only true gotcha is the API model-ID format — claude-opus-4-8, dashes, not a dot.

Once you’re on it, the changes worth knowing about — the 4× lower prompt-cache minimum, 3× cheaper fast mode, mid-conversation system messages, and Dynamic Workflows — are covered in Claude Opus 4.8: the 4 changes builders should actually care about.

Sources

• Claude Code model configuration — Claude Code docs
• Models overview — Anthropic docs
• Introducing Claude Opus 4.8 — Anthropic
• Claude Opus 4.8 is now available on AWS — AWS
• Claude Opus 4.8 is generally available for GitHub Copilot — GitHub Changelog
• Companion: Claude Opus 4.8 — the 4 changes builders should actually care about