How to Use Claude Mythos 5: Access and Practical Limits

Claude Mythos 5 is a model in the Claude lineup, available to users whose plan or API access includes it. Access may be limited at first, so the first step is always to confirm whether your account is eligible before trying to select the model. Once you have access, using it follows the same pattern as any other Claude model: pick it in the model selector or set it in your development environment.

Project context — the task notes, the decisions in play and where work was left — does not transfer automatically when you move between Claude models. Tools like fenbs keep that record outside any one model, so the context is ready whether you are on Mythos 5, an earlier Claude release, or a different assistant entirely.

Check whether your plan includes Mythos 5

Not every Claude plan includes every model, and a new model may roll out to plans in stages. Open the Anthropic pricing page↗ (opens in a new tab) and look for Mythos 5 in the list of models shown for your subscription tier. If the page shows Mythos 5 only under a tier you do not hold, you will need to upgrade or wait for wider availability.

In claude.ai, a model that is not available on your plan either does not appear in the model selector or appears grayed out with an upgrade prompt. In the Anthropic API, requesting a model ID that your key does not have access to returns an error rather than falling back silently to another model.

For a related model with different eligibility, see how to use Claude Fable 5. Both are part of the Claude 5 family but serve different use cases, and plan coverage may differ.

Select Mythos 5 in claude.ai

If your plan includes Mythos 5, open a new conversation in claude.ai and look for the model selector near the compose area. It shows the name of the model currently active. Click it and choose Mythos 5 from the list. The selection applies to that conversation; a new conversation may revert to your plan's default.

If you use Claude Projects, open the project settings and look for a default model option. Setting Mythos 5 there means new conversations in the project start with it pre-selected. Check this setting after plan changes, since updates can reset project defaults.

Select Mythos 5 in Claude Code

Claude Code uses the model set in the ANTHROPIC_MODEL environment variable. Set it before starting a session to use Mythos 5:

export ANTHROPIC_MODEL=claude-mythos-5-latest
claude

Use the exact model ID shown in the Anthropic API documentation↗ (opens in a new tab). Model IDs are versioned; latest aliases work but a pinned ID gives more reproducible results. Claude Code does not fall back silently if the ID is wrong or inaccessible — it returns an error.

You can add the ANTHROPIC_MODEL line to a .env file in your project directory so it loads automatically. Claude Code reads .env files when configured to do so, which saves resetting the variable each session.

Use Mythos 5 through the Anthropic API

If you call Claude through the API in your own code, pass the Mythos 5 model ID in the model field of each request:

{
  "model": "claude-mythos-5-latest",
  "max_tokens": 1024,
  "messages": [{ "role": "user", "content": "Your prompt here" }]
}

Check the Anthropic API documentation↗ (opens in a new tab) for the exact model ID and any rate limits specific to Mythos 5. Early-release models sometimes have lower rate limits that Anthropic raises over time. Use the API docs rather than guessing the model string, since an incorrect ID returns an error with no silent fallback.

What to carry between model sessions

When you switch from an earlier Claude model to Mythos 5, the new model starts with no memory of earlier sessions. Before switching, note what was completed, what was checked, what is still uncertain, and the decisions that continue to apply. A short Where we left off note in fenbs is the practical way to capture this.

At the end of a session on any model, ask the assistant to save progress in fenbs. At the start of the Mythos 5 session, ask it to read the project context. It gets the current task state, rules and next steps without you retyping anything.

How to do this with fenbs

1. Create a free account and add your project to fenbs.
2. Write the task description, any Decisions and rules that apply, and a Where we left off note covering the current state of the work.
3. Connect Claude to fenbs over MCP from Integrations. The same connection works across Claude models including Mythos 5.
4. At the start of each Mythos 5 session, ask Claude to read the project context from fenbs before you give the first prompt.
5. Before the session ends, ask Claude to save progress so the next session — on any model — starts from the current state.

For the broader Claude setup — plans, the web app and what to expect on first use — see what is Claude AI. For the API and how to authenticate, see the Claude API guide. For questions about Claude Code specifically, see what is Claude Code.