Short answer: You can no longer select Claude Opus 4.1 in Claude Code using a Claude subscription: Anthropic says it removed Opus 4 and 4.1 from the Claude model selector and Claude Code. (Claude release notes) The Anthropic API snapshot claude-opus-4-1-20250805 had a scheduled retirement date of August 5, 2026; that date has passed. (Anthropic API release notes) Anthropic says requests to retired models fail. (Model deprecations) The current opus alias resolves to Opus 5.5 on the Anthropic API. (Claude Code model configuration)
Amazon Bedrock is a separate, temporary possibility—not a guarantee that you can still use Opus 4.1 in Claude Code. AWS lists the model as Legacy in all Regions, with end of life scheduled for January 8, 2027. Under AWS’s lifecycle rules for models launched before September 7, 2026, new customers cannot use Legacy models, and existing customers may lose access after 15 days of inactivity. Claude Code supports Bedrock generally, but the documentation reviewed here does not confirm that Opus 4.1 can specifically be selected through Claude Code. (AWS model information)
Availability by channel
| Channel | Opus 4.1 status as of November 12, 2026 | Practical choice |
|---|---|---|
| Claude Code with a Claude subscription | Anthropic says Opus 4 and 4.1 were removed from the Claude model selector and Claude Code. | Select a model currently offered in Claude Code. (Claude release notes) |
| Direct Anthropic API | The claude-opus-4-1-20250805 snapshot’s scheduled retirement date was August 5, 2026. Anthropic says requests to retired models fail. (API retirement notice; model deprecations) |
Don’t build a new integration around that snapshot. Anthropic’s model configuration maps the opus alias to Opus 5.5. (Claude Code model configuration) |
| Claude Code through Amazon Bedrock | AWS lists Opus 4.1 as Legacy in all Regions, with end of life on January 8, 2027. AWS’s Legacy policy bars new customers from using Legacy models; existing customers may lose access after 15 days of inactivity. Claude Code supports Bedrock, but the cited Claude Code documentation does not specifically confirm Opus 4.1 selection. | Treat this as conditional, not a dependable route: confirm both model access in your AWS account and that Claude Code accepts the relevant Bedrock model configuration. Plan to migrate before the AWS end-of-life date. (AWS model information) |
| Other clouds, routers, or gateways | Availability and retirement depend on that provider. A third-party model listing alone does not prove it can still serve requests. | Check the provider’s current model catalog and lifecycle terms; don’t assume its model alias, snapshot, or billing matches Anthropic’s. |
Anthropic’s API retirement notice named Opus 4.8 as its migration recommendation at the time. (Anthropic API release notes) Its current model overview recommends Opus 5.5 as a starting point for most workloads. (Anthropic models overview) For a new Claude Code or API integration, follow current model guidance rather than treating an earlier migration notice as the latest recommendation.
Was Opus 4.1 good for Claude Code?
There is historical evidence that Opus 4.1 improved on Opus 4 for coding, but that does not make a retired model a sound choice for a new production integration.
Anthropic reported 74.5% on SWE-bench Verified for Opus 4.1 and specified that the result did not use extended thinking. Its release announcement also relayed reports from GitHub, Rakuten, and Windsurf about coding-related improvements. These are historical capability signals, not a guarantee that Opus 4.1—or a current alternative—will perform best on your repository. (Anthropic’s Opus 4.1 announcement)
An early LinkedIn post described improved large-codebase search and more precise edits, and said the author had fixed two bugs earlier models had missed. Those are the author’s impressions: the post gives no bug examples, prompts, comparison models, or controlled test method. Use them as ideas for what to evaluate, not as independent proof of general superiority in Claude Code.
A practical way to choose a replacement
Run a small, repeatable evaluation on tasks that resemble your real work before committing to a model or provider:
- Select a few representative issues from your repository—for example, a cross-file bug, a refactor with existing tests, and a change that requires investigation before editing.
- Give each candidate model the same starting commit, task description, tool permissions, and test commands.
- Compare whether the requested change works, whether tests and lint pass, whether unrelated files changed, how much human correction was needed, and the total usage cost and elapsed time.
- Prefer the model that meets your quality bar at an acceptable cost and workflow—not the one with the strongest result on a different benchmark.
This is a proposed evaluation method, not a test of Opus 5.5 or any other model. Claude Code’s opus alias resolves to Opus 5.5 on the Anthropic API; the model configuration differs by provider. (Claude Code model configuration)
API pricing: historical Opus 4.1 versus a current option
Anthropic’s pricing page lists Opus 4.1 rates, but its API retirement date has passed. Treat those rates as historical reference, not as confirmation that new Opus 4.1 API calls are available. The current table lists Opus 5.5 at $4 per million input tokens and $20 per million output tokens. (Anthropic API pricing)
| Model and pricing context | Input | Output |
|---|---|---|
| Opus 4.1, listed rates | $15 / million tokens | $75 / million tokens |
| Opus 5.5, current listed rates | $4 / million tokens | $20 / million tokens |
Illustrative price-only calculation: At 2 million input tokens and 200,000 output tokens, those rates work out to $45 at Opus 4.1’s listed rates versus $12 at Opus 5.5’s listed rates. This assumes the same token counts, excludes caching and other applicable charges, and is not a prediction of actual migration cost. Measure actual usage on representative tasks before estimating your project’s spend. (Anthropic API pricing)
API metering and Claude Code access are separate questions: the API pricing table is not a Claude Code subscription price. Claude Code supports different providers, including Claude subscriptions, Anthropic Console, and Amazon Bedrock; access and billing depend on the provider you use. Check the applicable terms for your route before estimating project cost. (Claude Code feature availability)
Bottom line
- Using Claude Code with a Claude subscription? Opus 4.1 has been removed from the model selector. Choose a currently listed model. (Claude release notes)
- Calling Anthropic’s API directly? Don’t build around the Opus 4.1 snapshot whose scheduled retirement date was August 5, 2026. Anthropic says requests to retired models fail; its current
opusalias maps to Opus 5.5. (Anthropic API release notes; model deprecations; Claude Code model configuration) - Considering Bedrock? AWS lists Opus 4.1 through January 8, 2027, but marks it Legacy. New customers cannot use Legacy models, and existing users may lose access after 15 days of inactivity. Verify AWS account access and Claude Code model selection before relying on this route. (AWS model information)
- Choosing based on coding quality? Opus 4.1 had positive historical coding results, but they don’t establish how a current model will perform on your code. Compare available models on a small, controlled set of your own repository tasks.
Sources
Claude Help Center release notes
Claude Platform release notes — Opus 4.1 retirement notice
Claude Code model configuration
Claude Opus 4.1 on Amazon Bedrock
Amazon Bedrock model lifecycle (Legacy)