Plans, limits & access
You don't need to memorize pricing; it changes often, but understanding how access works helps you avoid surprises.
The shape of the plans
Claude is available through a free tier and several paid tiers. As of mid-2026 the consumer tiers are roughly:
| Tier | Who it's for | What you generally get |
|---|---|---|
| Free | Trying Claude | Access to capable models, limited usage per period |
| Pro | Individual professionals | Higher limits, access to top models, Projects, more features |
| Max | Power users | Much higher usage, priority access to the newest models |
| Team | Small teams | Shared workspace, collaboration, central billing |
| Enterprise | Organizations | Admin controls, security, scale, data governance |
Details change
Exact prices, model availability, and limits shift frequently. Always confirm on Anthropic's official pricing page before deciding.
Usage limits, in plain terms
Paid plans give you a budget of usage over a rolling window (often a few hours), not unlimited use. A few things to know:
- Bigger models cost more of your budget per message. The most capable models (Opus and Mythos-class) consume usage faster than lighter ones (Haiku, Sonnet).
- Long conversations cost more. Every message re-reads the whole chat, so very long threads use more of your budget. Start fresh chats for new topics.
- Files and large context cost more. Uploading a 200-page PDF and asking about it uses far more than a one-line question.
The API is different
If you build with the API, you don't pay a subscription, you pay per token (input + output). This is usage-based and scales with how much you send and receive. See Choosing the right model for the cost/quality trade-off.
Practical tips to stretch your usage
- Use a lighter model for simple tasks (drafting, formatting, quick questions).
- Reserve top models for hard reasoning, important writing, and complex analysis.
- Trim context: don't paste a whole document when a section will do.
- Start new chats for unrelated tasks instead of one endless thread.
Now that access is clear, let's understand the different "brains" you can choose from: the Claude model family.