- Claude Subscription Plans at a Glance
- Claude Free Plan
- Claude Pro: $20/month
- Claude Max Plans: $100–$200/month
- Claude Team Plans: $25–$125/seat/month
- Claude Enterprise Plan
- Subscription vs API: Which Track to Choose
- How Claude Code Affects Your Plan Choice
- Token Costs: What the API Actually Charges
- Cost Reduction Paths
- Using Novita AI as an Alternative Backend
- FAQ
Claude subscription pricing in 2026 comes in two distinct tracks that confuse people constantly: a flat monthly subscription for using Claude through claude.ai and Claude Code, and a separate pay-as-you-go API for building applications. Picking the wrong track — or conflating the two — leads to real overpaying. This article covers every subscription tier, what you actually get from each, how subscription usage caps interact with Claude Code, and when token pricing through the API makes more financial sense.
Claude Subscription Plans at a Glance
| Plan | Monthly Price | Annual (per month) | Who It’s For |
|---|---|---|---|
| Free | $0 | $0 | Occasional use |
| Pro | $20 | ~$17 | Individual developers, regular users |
| Max 5x | $100 | — | Heavy Claude Code users |
| Max 20x | $200 | — | Full-day agentic workflows |
| Team Standard | $25/seat | ~$20/seat | Teams needing shared workspace |
| Team Premium | $125/seat | ~$100/seat | Engineering teams needing Claude Code |
| Enterprise | Custom | Custom | Large orgs, compliance requirements |
All prices are USD. Annual billing on Pro, Team Standard, and Team Premium saves approximately 15–20%.
Claude Free Plan
The free plan gives access to Claude’s web interface with limited usage. There is no Claude Code access, no API access, and no access to the full range of model tiers — Opus is gated behind paid plans. The free tier works for occasional questions or evaluating whether Claude fits your workflow. It is not a viable option for development, daily writing, or anything requiring consistent output.
Claude Pro: $20/month
Pro is the entry point for serious use. The key changes over Free:
- 5x more usage than the free tier
- All current model tiers — Haiku 4.5, Sonnet 4.6, and Opus 4.8
- Claude Code included — Anthropic’s terminal-based coding agent
- Research mode — extended web research tasks
- Unlimited projects — organized workspaces with persistent context
- Microsoft 365 integrations
At ~$17/month on the annual plan, Pro is the most cost-effective entry point for anyone using Claude as a regular tool. The main limitation: Claude Code and multi-file coding sessions consume usage allowances significantly faster than conversational tasks. If you run Claude Code for several hours daily, you will likely hit the Pro ceiling.
Claude Max Plans: $100–$200/month
The Max tiers exist specifically because Claude Code exhausts usage limits fast. An extended coding session — reading a codebase, revising files, running checks — consumes usage at roughly 5–20x the rate of a chat conversation.
Max 5x ($100/month): Five times Pro’s usage ceiling. Suited for developers who use Claude Code as a daily coding tool but not as their exclusive development environment throughout the workday.
Max 20x ($200/month): Twenty times Pro’s ceiling. Targets developers who run Claude Code as their primary IDE assistant across full working days, or researchers and agents running extended multi-step tasks. This tier also includes early access to new features before general availability.
Neither Max plan currently offers annual billing. If you find yourself regularly hitting Pro limits within a week, Max 5x is the practical next step. Max 20x only makes sense if Max 5x is still insufficient.
Claude Team Plans: $25–$125/seat/month
Team plans add shared administration on top of individual plan features.
Team Standard ($25/seat/month, ~$20/seat annual):
- Pro-level model access and usage per seat
- Shared workspace and admin controls
- SSO and central billing
- Desktop app deployment management
This plan suits organizations that want team coordination, usage visibility, and identity management without requiring Claude Code for every seat. Minimum seat counts may apply.
Team Premium ($125/seat/month, ~$100/seat annual):
- 5x usage per seat — equivalent to an individual Max 5x
- Claude Code access for every seat
- Same admin and SSO features as Standard
- Minimum 5 seats
Team Premium is the right call when engineers actively use Claude Code. The per-seat annual rate (~$100) is roughly equivalent to buying each engineer a Max 5x individual plan, but with centralized admin and billing.
Claude Enterprise Plan
Enterprise pricing is custom. The self-serve entry starts around $20/seat plus API usage, but actual terms depend on contract size and requirements. Enterprise contracts typically include:
- Custom rate limits
- Volume discounts
- Dedicated support
- Advanced security and compliance: SSO, audit logs, data residency options
- Negotiated per-seat and API rate caps
If you are evaluating enterprise primarily for API access, it is worth separating the subscription cost from API token cost in contract negotiations — they are billed independently and optimized differently.
Subscription vs API: Which Track to Choose
The most common mistake: paying for a Pro or Max subscription and using it as a proxy for API access. The subscription controls access through the claude.ai web interface and Claude Code — it does not grant unlimited API token usage. API calls are billed separately at per-token rates.
Use a subscription when:
- You use Claude primarily through the web interface or Claude Code
- You want a predictable monthly cost without tracking token usage
- You need Research mode, projects, or Microsoft 365 integrations
Use the API directly when:
- You are building an application that calls Claude programmatically
- You need cost control at the token level
- You want to use Batch API discounts or prompt caching
- Your volume makes per-token pricing cheaper than a subscription ceiling
For developers building products: buy API credits, pick the cheapest model that passes quality bar for your specific task, and use the Batch API for non-realtime jobs.
How Claude Code Affects Your Plan Choice
Claude Code’s usage model is the primary reason Max plans exist. A few concrete benchmarks:
- Simple chat conversation: 1x usage consumption
- Single-file code edit: 3–5x faster consumption than chat
- Multi-file refactor or agentic session reading a codebase: 10–20x faster consumption
If you use Claude Code for 1–2 hours of focused work per day, Pro ($20/month) often holds. If Claude Code is running for most of your workday across multiple sessions, Max 5x ($100/month) is the realistic floor. Full-day heavy usage across multiple projects typically requires Max 20x ($200/month).
The practical test: start on Pro for one week and monitor whether you hit limits. If you hit them by day 3, upgrade to Max 5x. If you hit Max 5x limits regularly, Max 20x is warranted.
Token Costs: What the API Actually Charges
For developers building on the API, Claude’s per-token pricing as of July 2026:
| Model | Input ($/MTok) | Output ($/MTok) |
|---|---|---|
| Claude Opus 4.7 / 4.8 | $5.00 | $25.00 |
| Claude Sonnet 4.6 | $3.00 | $15.00 |
| Claude Sonnet 5 | $2.00* | $10.00* |
| Claude Haiku 4.5 | $1.00 | $5.00 |
*Claude Sonnet 5 introductory pricing through August 31, 2026; standard rate of $3/$15 takes effect September 1, 2026.
Output tokens cost 5x the input rate on every tier — a consistent pattern that matters for any completion-heavy workload.
Model choice is the biggest cost lever. For classification, extraction, or summarization tasks, Haiku 4.5 at $1/$5 per MTok covers most workloads at 80% lower cost than Opus. Test cheaper models first before committing a workload to Opus.
Cost Reduction Paths
Prompt caching: For repeated context — system prompts, reference documents, codebase snapshots — prompt caching reuses processed tokens at 10% of the standard input price on cache reads. Cache write has a small surcharge (1.25x for 5-minute TTL, 2x for 1-hour TTL). The economics pay off after one or two cache reads.
Batch API: Asynchronous processing with 50% off both input and output tokens. Returns results within 24 hours. Direct fit for eval pipelines, bulk content generation, nightly data processing, or any workload where real-time latency is not required.
Combined: a pipeline that reuses a large system prompt and batches requests can reduce effective per-token cost by 70–80% versus synchronous non-cached calls.
Using Novita AI as an Alternative Backend
For developers building cost-sensitive applications or workflows that don’t strictly require Claude, Novita AI LLM API provides OpenAI-compatible endpoints for open-source models including Llama, Qwen, DeepSeek, and GLM at substantially lower token rates — starting at $0.02/million tokens for smaller models and under $0.50/million tokens for capable mid-size models.
Several of these models work directly as Claude Code backends via the API compatibility layer. GLM-5, for example, is priced at $1.00/$3.20 per million tokens on Novita AI and has been tested for agentic coding workflows — a fraction of Opus token rates. See how to use GLM-5 in Claude Code for setup details.
For teams running multi-step agentic workflows, Novita Agent Sandbox provides managed execution environments for browser automation, code execution, and file operations. For a full breakdown of sandbox infrastructure costs, see AI agent sandbox pricing models.
If budget is the primary driver and Claude is not a hard requirement, see top 10 cheapest LLM APIs in 2026 for a cross-provider ranked comparison.
FAQ
What is Claude Pro price?
Claude Pro costs $20/month billed monthly, or approximately $17/month on an annual plan.
How much does Claude Max cost?
Claude Max 5x is $100/month. Claude Max 20x is $200/month. Neither currently offers annual billing.
What is Claude Code cost by plan?
Claude Code is included with Pro ($20/month) and above. Free plan has no Claude Code access. Team Premium ($125/seat/month) is the team-level equivalent that includes Claude Code for every seat.
How much does Claude Team plan cost?
Team Standard is $25/seat/month ($20/seat annual). Team Premium is $125/seat/month ($100/seat annual). Minimum seat counts may apply.
What is Claude Enterprise plan pricing?
Enterprise is custom-priced. A self-serve entry starts around $20/seat plus API usage costs. Larger contracts include volume discounts, dedicated support, and compliance features.
What is the difference between a Claude subscription and API access?
A subscription controls usage through the claude.ai web interface and Claude Code. API access is billed separately per token, regardless of your subscription tier. The two are independent billing tracks.
How much does the Claude API cost?
As of July 2026: Haiku 4.5 at $1/$5 per million tokens (input/output), Sonnet 4.6 at $3/$15, Opus 4.7/4.8 at $5/$25. Sonnet 5 at introductory $2/$10 through August 31, 2026. All rates are pay-as-you-go.
Does Anthropic offer a discount on Claude plans?
Annual billing reduces monthly rates by approximately 15–20% on Pro, Team Standard, and Team Premium. The API offers the Batch API (50% off both token types) and prompt caching (90% off repeated context reads) as structural discount mechanisms.
