To switch Claude models in a project, open the project, click the model name under the message box (for example Opus 5 Medium) and pick another model, such as Sonnet 5.5 or Opus 5.5. There is no separate model setting on the project page: you choose the model for each chat, and you can change it again in the middle of a chat.
This guide is for anyone who works in Claude projects on claude.ai in the browser and wants a faster, cheaper or more capable model for a given task. You need a Claude account with at least one project. On Enterprise, an admin may have disabled some models or capped the effort level for your role.
How to switch Claude models in projects
- Open Projects. Click Projects in the left sidebar. The Projects page lists your projects, with tabs for Your projects, Organization and Shared with you.
-
Open the project. Click the project’s card. The project page shows its name and goal at the top, a message box on the left, and the Instructions, Files, Memory and Scheduled rows on the right.
-
Click the model name under the message box. Below the project’s message box, next to Auto, you see the current model and effort level, for example Opus 5 Medium. Click it to open the model menu.
- Pick a model. The menu lists the main models with a one-line hint each: Fable 5.1 (for your toughest challenges), Opus 5.5 (for complex work and everyday tasks), Sonnet 5.5 (most efficient for simpler tasks) and Haiku 5.5 (fastest for quick answers). Click one. The selector now shows the new name, and your next message goes to that model.
- Use More models for a specific version. If your team has settled on one version, hover More models. The submenu lists other versions, such as Mythos 5.1, Fable 5, Opus 5, Opus 4.8 and Sonnet 5. A check mark shows the current choice. The menu is the same as on the home screen.
- Set the effort level if needed. Hover Effort in the same menu and choose Low, Medium, High, Extra or Max. The menu explains that higher effort gives more thorough answers but takes longer and uses your limits faster. Max is flagged as 1.5x or more usage.
- Send your message, or switch inside a running chat. Type your request and press Enter. The chat opens with the project’s instructions and files loaded. In a chat that is already open, the model name sits below the Reply box, on the right. Click it to switch at any point. The change applies from Claude’s next response.
Problems when switching models in a project
- You are looking for a project-wide setting. The project page has no default-model option. You pick the model in each chat with the menu under the message box. On Enterprise, Anthropic’s admin documentation says Claude remembers the model each person last picked in each project.
- Every new chat starts on the same model. On Enterprise, an owner can set a default model under Organization settings > Models and turn on “Always start with the default model and effort level” (beta). With that on, new projects start on the default model, while existing chat projects start on the model you last chose in that project. You can still switch inside any chat.
- A model or effort level is missing from the menu. On Enterprise, your admin may have disabled some models or capped the effort level for your role. Ask an owner which models are enabled.
- The answer did not change after you switched. The switch applies from the next response. Earlier replies in the chat stay as they were. Send a new message, or start a fresh chat in the project if you want the whole thread on one model.
- Teammates see a different model in the same project. That is expected. The model choice is per person and per chat, not stored with the project’s instructions or files.
Switching models in Claude Code and Cowork
Claude Code. In the terminal, you switch models with the /model command or the --model flag at launch, not through the web menu. See how to change model in Claude Code.
Cowork. Anthropic states that projects work the same way in Cowork and in chat, and you can import a chat project into Cowork with its files and instructions.
New version of projects. Anthropic is rolling out a new version of projects in stages, starting with Claude Code, where a project is one conversation that Claude splits into parallel threads. If your project page looks different, follow what you see on screen.
Rolling out model choice across a team
When 20 people pick models on their own, three things go wrong. Some run every small task on the largest model at Max effort, which burns through usage limits fast. Others stay on the fastest model for work that needs depth, then blame Claude for weak answers. And because the model lives in each chat, two people in the same shared project can get different results from the same instructions and files, which makes quality hard to review.
What to agree on:
- A default per type of work. For example, a fast model for quick rewrites and summaries, a larger one for analysis and drafting. Write it into the project’s instructions or your team guide so people do not guess.
- When to raise effort. Medium for daily work, higher for hard reasoning, Max only when the task justifies the extra usage.
- Admin defaults on Enterprise. Set the organization default model, decide whether “Always start with the default model and effort level” should be on, and limit models per custom role where cost control matters.
- One review habit. When output quality varies, check the model and effort level in the chat first, before rewriting instructions.
For help on picking between the main models, see how to use Claude Fable. For what a project is and how to set one up, see how to use Claude projects.
Frequently asked questions
Can I set a default model for a Claude project?
Not from the project page. You choose the model per chat with the menu under the message box. On Enterprise, Claude remembers the model each person last picked in each project, and an owner can set an organization default model under Organization settings > Models.
Can I change the model in the middle of a project chat?
Click the model name below the Reply box and pick another model. The change applies from Claude’s next response, and the project’s instructions and files stay loaded.
Does switching models change my project’s instructions or files?
Switching models leaves instructions and files untouched. They belong to the project and load in every chat, whatever model you pick. Only the model that writes the next answer changes.
Why can’t I see a certain model in my project?
The menu shows the models available to your account. On Enterprise, an admin may have disabled some models or capped the effort level for your role. Check More models first, since some versions only appear there.
Train your whole team on Claude
Your team is figuring this out one at a time. We train them together in a day.
