At some point while using Claude, you may see a notice that a model you rely on is being deprecated (deprecation — when a specific model is no longer offered). Based on Anthropic's official Claude Help Center article, this guide walks through how to adapt smoothly to a new model before the one you're used to disappears. Every step is explained in plain language, so no technical background is needed.
🟢 Current lineup (September 2026): Claude Opus 5.5 / Claude Sonnet 5 / Claude Haiku 4.5 (higher tier: Claude Fable 5.1). · model notice · Fable subscription
Fable 5 and 5.1 subscription (updated September 7, 2026): Claude Fable 5.1, released September 1, 2026, is the current Fable model and Fable 5 is now legacy. Plan terms are the same for both — Max and Team Premium plans include Fable at up to 50% of the weekly usage limit; Pro and Team Standard use usage credits (
What Is Model Deprecation?
Model deprecation is when a specific Claude model is no longer available to use — a routine part of a model's normal lifecycle (the process from release to eventual retirement), according to the official documentation. Anthropic states it would like to keep past models publicly available, but maintenance overhead and capacity constraints don't currently allow this as it continues to advance safety and capabilities. At the same time, it recognizes that users who value a specific model's unique character or capabilities on a personal level bear real costs when that model goes away, and says it aims to provide resources, tools, and guidance to make transitions as smooth as possible.
What to Check Before You Start
Before you start switching over, check these two things.
- Confirm the deprecation notice — Check whether the model you're using has received a deprecation notice and how long it remains available. For the exact timeline and model name, check the notice on your screen or the official release notes directly.
- List the conversations and prompts you rely on most — Jot down a few conversations or prompts (instructions) that have mattered most to you. You'll use these as a benchmark when comparing against the new model in the next step.
Quick Terms to Know
| Term | Plain-language meaning |
|---|---|
| Deprecation | Discontinuing a specific model |
| Model lifecycle | The full process a model goes through, from release to eventual retirement |
| Memory | A feature that lets Claude remember past conversations and search them for reference |
| Projects | A workspace in Claude where you can store custom instructions and reference material (context) ahead of time |
| Custom instructions | Instructions you set in advance to tell Claude "respond in this tone, in this way" |
How to Adapt to a New Model — Official Recommended Strategies
The official documentation says combining the strategies below helps in most situations. Not every strategy fits every situation, so it's worth experimenting to find the combination that works for you.
- As soon as a deprecation notice appears, start experimenting with the new model. Run the same prompts and conversations that matter most to you through the new model, and compare it side by side with the current one. This lets you find the right alternative before the old model is actually retired.
- Turn on memory and let Claude search your past conversations. The exact setting location should be confirmed on screen, but having this on helps a new model pick up your tone and preferences faster.
- Save your personalization in Projects. A project can hold custom instructions and frequently referenced material (context) ahead of time, which helps a new model understand and follow what you need.
- Give direct feedback during conversations to shape the model's response style. If a new model feels more or less talkative than you'd like, or its tone feels off, tell it what you'd prefer — within that conversation, at the project level, or across your whole account. The official documentation describes this personalization work itself as "giving feedback to new models."
- If you can't put the style you want into words, ask Claude to figure it out. The official documentation suggests asking Claude to review the conversations you've most appreciated and craft its own instructions based on what it finds.
New Models Bring Changes Too — Worth a Look
Rather than focusing only on reproducing an old model's style, it's worth looking at what a new model does better. According to the official documentation, newer models often show improvements in performance on complex topics, brainstorming and collaboration, and instruction following (accurately doing what you asked). For example, as of this article's latest update (August 2026), Claude's top-tier model is Opus 5, its mid-tier model is Sonnet 5, and its lightweight model is Haiku 4.5. The exact current lineup can change over time, so it's safest to check the official models page for the latest details.
Common Sticking Points and Fixes
- "I turned on memory but the new model still doesn't sound like the old one." — Memory only helps a model reference past conversations; it doesn't fully replicate an old model's personality. The official documentation itself states plainly that these strategies aren't perfect and can only go so far toward preserving or replicating the unique experience of a particular model. Combining in-chat feedback with Project instructions helps narrow the gap.
- "I don't know how to describe the style I want." — As described above, ask Claude to review the past conversations you've been most satisfied with and to draft its own instructions based on those.
- "I don't know the deprecation timeline or the exact model name." — Exact timelines and model names change independently of the general principles in this article, so check the notice on your Claude screen or the official release notes directly for accuracy.
Not a Perfect Fix — Worth Knowing
These strategies make the transition smoother, but they can't fully bring back the exact experience you had with a specific model. The official documentation acknowledges this, while also noting that Anthropic is already taking early steps toward model preservation and aims to someday make past models publicly available again in some form. As a general reference point, it also notes that Anthropic has historically tended to deprecate most models around a year after their initial release.
Frequently Asked Questions
Q. Can I use a deprecated model again later?
There's currently no documented way to immediately use a model again right after it's deprecated. However, the official documentation states that Anthropic is taking early steps toward model preservation and aims to someday make past models publicly available again in some form.
Q. Do I need to use both memory and Projects?
No, using just one still helps. That said, the official documentation recommends combining several strategies, including these two, and suggests experimenting to find which combination works best for your situation.
Q. How long does it usually take before a model is deprecated?
According to the official documentation, Anthropic has historically tended to deprecate most models around a year after their initial release. This is a general pattern only — check the notice for your specific model for the exact timeline.
Next Steps
Even if the model you're using now has no deprecation notice, turning on memory and organizing your frequently used instructions in a Project ahead of time will make any future transition much easier. It also helps to get in the habit of noting which conversations you find most satisfying.
Related articles
- How to Change Model, Effort & Extended Thinking in the Claude App
- Claude Projects, Artifacts & Memory: What They Are and How to Use Them
- How to Write Prompts for Claude: 6 Core Principles from the Official Docs
- How to Use Claude Projects: Knowledge & Instructions to Cut Repetition
- Claude Artifacts: How to Create, Edit, Publish & Share (2026)