Full breakdown of Claude Code subscription plans in 2026: Pro ($20/mo), Max ($100/mo, $200/mo), Team, and Enterprise. What's included, usage limits, and how to pick the right plan.
Claude Code is included in Claude's subscription plans — you don't need a separate API key or usage-based billing to run it. Here's how the plans break down for engineering use cases.
| Plan | Price | Claude Code access | Usage limits | Best for |
|---|---|---|---|---|
| Claude Pro | $20/mo | Yes | Moderate (limited heavy agentic sessions) | Developers using Claude Code for a few hours per week |
| Claude Max (5× limits) | $100/mo | Yes, unlimited | 5× Pro limits | Engineers using Claude Code as a daily driver |
| Claude Max (20× limits) | $200/mo | Yes, unlimited | 20× Pro limits, priority access | Power users, AI-first engineering teams |
| Claude Team | $30/user/mo (min 5) | Yes, per seat | Higher than Pro per seat | Engineering teams needing admin, SSO, shared projects |
| Claude Enterprise | Custom | Yes, all seats | Negotiated, highest | Large orgs with compliance, custom data retention needs |
Claude Code uses tokens — the same resource as claude.ai chat. Every message you send, every file Claude reads, and every code block Claude generates consumes tokens from your plan's monthly allowance. Usage resets monthly.
| Activity | Typical token cost | Notes |
|---|---|---|
| Simple one-shot question | 500–2K tokens | Quick fixes, explanations |
| Multi-file refactor session | 20K–100K tokens | Reads multiple files + writes diffs |
| Full agentic PR workflow | 100K–500K tokens | Read repo → write tests → open PR → iterate |
| Large context session (200K window) | 200K+ tokens | Loading entire codebase context |
On Claude Pro, most developers find they hit limits within 2–3 hours of intensive agentic work. Signs you need to upgrade to Max:
| Scenario | Cheaper option | Why |
|---|---|---|
| Occasional use (1–2 sessions/week) | API (pay-per-token) | Flat $100/mo subscription wastes money if you use <$100 of tokens |
| Daily driver (1–4h/day) | Max plan ($100/mo) | Sonnet 4.6 at $3/1M input tokens; 100K tokens/session = $0.30/session → $90+/mo on API |
| CI pipelines (always-on) | API with prompt caching | Cached reads at $0.30/1M tokens; caching + Haiku slashes cost; subscription doesn't cover API calls |
| Team of 5+ | Claude Team ($30/seat) | Per-seat is cheaper than 5× individual Max plans; includes admin dashboard |
# Set ANTHROPIC_API_KEY instead of logging in with claude login
export ANTHROPIC_API_KEY=sk-ant-api03-...
# Claude Code automatically uses API key if set
claude
# Or pass a model flag to use a cheaper model for lightweight tasks
claude --model claude-haiku-4-5-20251001When using API billing, your token usage appears in the Anthropic console under Usage. Use the Claude API Cost Calculator to estimate costs before committing to a workflow. For batch/async tasks, see the Batch API guide (50% cost reduction for non-realtime work).
For a cost comparison across Claude models, see the Claude API pricing guide. For headless Claude Code usage in CI (which requires API billing, not subscription), see the headless mode guide.