Yes. Claude Opus 4.5 is listed as available in Claude Code. To select it in a running session, enter /model, choose Opus 4.5, and check the active model with /status. To start a terminal session with it, run claude --model claude-opus-4-5-20251101. Anthropic’s Help Center lists that model ID and selection path. (Claude Code model configuration)
One qualification matters before building around it: Anthropic’s model reference labels Opus 4.5 a legacy model and says it will not retire before November 24, 2026; that is a minimum date, not a confirmed retirement date. Anthropic also labels Sonnet 4.5 deprecated and lists its retirement date as November 30, 2026. Check the models available to your account and their current lifecycle status before choosing either for a new production workflow. (Opus 4.5 model reference; Sonnet 4.5 model reference)
How to select Opus 4.5 in Claude Code
In a session that is already running, enter:
/model
Choose Opus 4.5 from the interactive menu. To verify the active model, run:
/status
To start a terminal session with Opus 4.5, use its explicit model ID:
claude --model claude-opus-4-5-20251101
Anthropic documents both the in-session menu and the --model option. Which models are available can depend on how you sign in and your organization’s settings; the /model menu shows the models available to your account. (Claude Code model configuration)
Opus 4.5 or Sonnet 4.5 for coding?
A qualified starting point, based on Codecademy’s model-selection guidance: try Sonnet 4.5 for routine, clearly scoped coding where speed or budget matters. Consider Opus 4.5 for complex debugging, multi-file work, or tasks requiring more planning—especially if Sonnet 4.5 needs repeated corrections or the task must be split into smaller steps. This is general advice, not a guarantee that one model will do better on your codebase. Test both on representative repository tasks when the choice matters. (Codecademy’s Opus 4.5 guide)
For direct Claude API use, Anthropic lists Opus 4.5 at $5 per million input tokens and $25 per million output tokens; Sonnet 4.5 is listed at $3 per million input tokens and $15 per million output tokens. These are API token rates, not flat Claude Code session fees. Actual cost for a task also depends on tokens used and applicable billing arrangements, so compare total usage and results on representative work rather than assuming the higher unit rate always means a higher cost per completed task. (Anthropic pricing)
Anthropic reported that, at medium effort on SWE-bench Verified, Opus 4.5 matched Sonnet 4.5’s best score while using 76% fewer output tokens. Treat this as Anthropic’s benchmark result under that stated setting—not a prediction of performance or cost on your codebase. (Anthropic’s Opus 4.5 announcement)
Claude Code or an API for a production application?
Use Claude Code when you want a coding tool to work with a software project. Use a model API when your application needs to send model requests and handle responses as part of its own functionality. For a direct Claude API integration, Anthropic identifies claude-opus-4-5-20251101 as the Opus 4.5 model ID; consult its API documentation for implementation details. Vertex AI and other cloud providers are separate access routes with their own setup and billing documentation. (Opus 4.5 model reference)
Bottom line
Select Opus 4.5 in Claude Code with /model, or start a terminal session using claude --model claude-opus-4-5-20251101. For the version-specific comparison, consider Sonnet 4.5 for routine work and Opus 4.5 for more complex tasks, but validate the choice on your own code. Their direct API rates are $3/$15 and $5/$25 per million input/output tokens, respectively; compare total usage and results rather than treating those rates as a guaranteed difference in task cost. Check lifecycle status before selecting either for a new production dependency.
Sources
Claude Code model configuration — Claude Help Center
Claude Opus 4.5 — Claude Platform Docs
Claude Sonnet 4.5 — Claude Platform Docs
Claude Opus 4.5 Tutorial for AI Agents and Coding — Codecademy