How much does Claude Enterprise cost per user?

Claude Enterprise charges $20/month for technical seats (Claude Code access) and $10/month for business seats (Claude.ai chat only). These are access fees only — all token usage is metered and billed separately at standard API rates on top of the seat cost.

What's the difference between technical and business seats?

Technical seats ($20/mo) include access to Claude Code, the CLI-based agentic development tool, plus Claude.ai chat and Cowork. Business seats ($10/mo) provide access to Claude.ai chat and document analysis only, without Code access.

How does Claude's token-based pricing work?

Every interaction is metered by tokens (roughly ¾ of a word). Input tokens (what you send) and output tokens (what Claude generates) are priced per million tokens, with output costing 3-5x more than input. Rates vary by model: Opus 4.6 ($5/$25), Sonnet 4.6 ($3/$15), Haiku 4.5 ($1/$5) per million input/output tokens.

Does Anthropic offer volume discounts for Claude Enterprise?

Under the 2026 pricing model, Anthropic has eliminated automated volume discounts. Committing to higher spend does not reduce the per-token rate. Organizations can negotiate roll-over credits, flexible true-up windows, and phased seat ramping, but the base token unit cost remains fixed.

How Much Will Claude Enterprise Cost Your Team?

Anthropic's 2026 pricing unbundles seat access from token consumption. Your actual cost depends on team size, usage intensity, and model mix. Use our calculator to estimate your annual budget.

↓

The New Pricing Model

Claude Enterprise now charges a flat monthly access fee per seat — $20 for technical users (Claude Code, Cowork, Chat) and $10 for business users (Chat only). But these seats include zero token usage. Every token processed is metered and billed at standard API rates on top of the seat cost.

Organizations must also commit to a mandatory pre-paid consumption minimum to unlock enterprise features like SSO, SCIM, and extended context windows. There are no automatic volume discounts — the per-token price stays fixed regardless of commitment size.

This makes budgeting non-trivial. A team of 50 developers could spend anywhere from $50,000 to $500,000+ annually depending on how they use Claude. The calculator below helps you model your specific scenario.

Team Composition

Light Technical

$20/seat/mo · ~25K tokens/day

Medium Technical

$20/seat/mo · ~200K tokens/day

Heavy Technical

$20/seat/mo · ~1M tokens/day

Business (Non-technical)

$10/seat/mo · ~5K tokens/day

Model Mix (% Sonnet / Opus / Haiku)

Light / / = 100%

Medium / / = 100%

Heavy / / = 100%

Business / / = 100%

Other Assumptions

Prompt Cache Hit Rate 75%

Working Days / Month

Daily Token Consumption (override)

Light (tokens/day)

Medium (tokens/day)

Heavy (tokens/day)

Business (tokens/day)

Estimated Cost

Add team members above to see your estimate

Optimize Your Claude Spend

Prompt Caching

Cache static prefixes (system prompts, tool definitions, file structures) to achieve a 90% reduction on repeated input tokens. Keep sessions warm — a cache miss after idle periods forces expensive context rebuilds.

Model Routing

Route orchestrator agents and simple tasks to Sonnet or Haiku. Reserve Opus for complex reasoning and architectural design. This alone can cut heavy-user costs by 60% or more.

Batch API

Use the Batch API for async workloads like code audits and automated PR reviews. It offers a 50% discount on all token rates in exchange for a 24-hour processing window.

Context Hygiene

Use .claudeignore to exclude build directories, binaries, and irrelevant files from context. Output tokens cost 3-5x more than input — instruct agents to generate minimal diffs, not full rewrites.

Need Help Optimizing Your Claude Deployment?

Marvik is an Anthropic partner with 8 years exclusively focused on AI. We've scaled production AI systems for companies like Arm, Apple, and Microsoft — we go beyond POCs.

Talk to Our Team