Claude AI Pricing 2026: Pro $20/mo, Max $100-$200, and Opus 5 API Costs

Vendor information last verified: 18 September 2026

Disclosure: ScreenApp sells AI transcription, meeting, recording, and video-analysis software and may compete with products discussed in this article. ScreenApp does not receive payment for rankings. Product claims, pricing, and specifications are verified against the sources listed in this article.

How this article was researched

ScreenApp did not perform first-hand product testing for this article. Product information was verified against the cited primary sources: Anthropic’s Claude pricing page, the Claude API pricing docs, the models overview, Claude Help Center articles, and Anthropic news posts. Each time-sensitive claim links to its source.

Not personally tested:

• The Free, Pro, Max 5x, Max 20x, Team, and Enterprise plans
• The models (Fable 5.1, Opus 5, Sonnet 5, Haiku 4.5, and older versions)
• Features such as Claude Code, Cowork, Projects, Research, memory, and connectors
• The API cost scenarios in this article, which are arithmetic on published prices, not measured bills
• Enterprise deployment, admin, and compliance features

Vendor-supplied information: Prices, usage multiples, model specs, and feature availability come from Anthropic. We have not independently measured usage limits, speed, or output quality. Anthropic says prices and plans can change at its discretion, so check the linked pages before you buy.

Anthropic’s current default model for most work is Claude Opus 5, with Claude Fable 5.1 above it for the most demanding reasoning and long-running agentic work, according to Anthropic’s models overview. Subscriptions run from free to $200/month.

Claude handles long documents (up to 1M tokens of context on current models), coding, and writing. This guide sticks to what Claude costs and what each plan includes.

How Much Does Claude Cost? (2026)

Claude costs $0 (Free) to $200/month (Max 20x) for individuals. Team seats start at $20/month per user on annual billing, and Enterprise is $20 per seat per month plus usage at API rates (Claude pricing).

Current Subscription Pricing:

• Free: $0. Sonnet and Haiku models, no Opus (source)
• Pro: $20/month, or $17/month on the annual plan ($200 billed up front). At least 5x the Free plan’s usage per 5-hour session, plus Claude Code and Opus (source)
• Max 5x: $100/month, 5x Pro’s usage per session (source)
• Max 20x: $200/month, 20x Pro’s usage per session (source)
• Team: Standard seat $20/month annual ($25 monthly), Premium seat $100/month annual ($125 monthly), for teams of 2 to 150 (source)
• Enterprise: $20/seat/month billed annually, plus usage at API rates (source)

Claude API Pricing (pay per use, per million tokens, input/output) (API pricing):

• Haiku 4.5: $1 / $5, the fastest and cheapest
• Sonnet 5: $2 / $10, speed and intelligence balanced
• Opus 5: $5 / $25, Anthropic’s recommended starting point for most workloads
• Fable 5.1: $10 / $50, for demanding reasoning and long-horizon agentic work

Batch API discount: 50% off input and output tokens for asynchronous jobs, so Sonnet 5 drops to $1/$5 and Opus 5 to $2.50/$12.50 (source).

Editorial judgement, not a test result. Criteria: how many hours a day you use Claude, whether you need Claude Code or Opus, and whether your usage is steady or spiky. On those criteria, Pro is the sensible default for people who use Claude most working days. Developers with occasional or automated workloads may pay less on the API, and Max only makes sense if you keep hitting Pro’s limits.

What is Claude AI?

Claude is Anthropic’s AI assistant, trained with Anthropic’s “Constitutional AI” approach to be helpful, harmless, and honest. The current lineup is Fable 5.1, Opus 5, Sonnet 5, and Haiku 4.5 (models overview).

What Anthropic lists for the current models:

• 1M token context window on Fable 5.1, Opus 5, and Sonnet 5; 200K on Haiku 4.5 (source)
• 128K max output tokens on Fable 5.1, Opus 5, and Sonnet 5; 64K on Haiku 4.5 (source)
• Adaptive thinking: Claude decides how much to reason, steered by an effort setting. On Fable 5.1 it is always on (source)
• Fast mode: up to 2.5x higher output speed on Opus 5 and Opus 4.8, at premium pricing, in research preview (source)
• Document analysis: PDF processing, contract review, and meeting transcript summarization
• Coding: Claude Code is included in all paid plans (source)

For businesses, Claude’s document handling suits AI meeting transcription, automatic meeting summaries, and AI note-taking from recorded conversations, then turning the results into shareable summaries. For media work specifically, see whether Claude can watch videos, how to chat with a YouTube video, and how to turn any video into text.

Note: Anthropic’s lineup changes often. Older models still available on the API include Fable 5, Opus 4.8, Opus 4.7, Opus 4.6, Opus 4.5, Sonnet 4.6, and Sonnet 4.5 (models overview).

Claude Models Explained

Anthropic released Sonnet 5 on June 30, 2026 (announcement) and Opus 5 on July 24, 2026 (announcement). Fable 5.1 is the successor to Fable 5 at the same input and output prices, with cheaper cache reads (what’s new in Fable 5.1).

Model Comparison

All figures below are from Anthropic’s models overview and plan comparison, checked 18 September 2026.

Fable plan terms are from Anthropic’s Fable models on your plan article. Sonnet 5 is the default model on Free and Pro (source), and Opus 5 is the default on Max (source). Haiku 4.5 has a retirement date of not sooner than October 15, 2026 (source).

Choosing a model (editorial judgement; criteria: task difficulty, cost per token, and speed):

• Sonnet 5 suits most everyday writing, documentation, code, and PDF work, and it is the cheapest current model with a 1M context window.
• Opus 5 is Anthropic’s recommended starting point for most API workloads and the step up for hard problems, large codebases, and agentic coding.
• Fable 5.1 is the top tier, at double the Opus token price. Anthropic suggests it when Opus 5 at higher effort still falls short.
• Haiku 4.5 is the speed-and-cost option for chatbots and simple, high-volume calls.

For AI-powered meeting analysis, a mid-tier model is usually enough for standard meetings, and a larger model is worth it for long multi-party calls. For tools built on models like these, see our guide to the best AI note-takers.

Claude AI Plans: Complete 2026 Pricing Breakdown

How usage limits work

Anthropic does not publish a fixed message count for any plan. Every plan has usage limits that reset on a rolling five-hour session window, and paid plans add weekly limits on top. Chats and Claude Code draw from the same pool, and how far a limit goes depends on conversation length, the model, and the features you use (Claude pricing FAQ). Paid plans can turn on usage credits to keep working at standard API rates after hitting a limit (source).

Free Plan: Getting Started

Cost: $0 (source)

Plan features (plan comparison):

• Sonnet and Haiku models, no Opus or Fable
• Chat on web, iOS, Android, and desktop
• Web search, file creation with code execution, Artifacts, memory, and connectors
• Up to 5 Projects
• No Claude Code, no Research, no Cowork (Cowork is paid plans only)

Best for (editorial judgement): Criteria: light or occasional use and no need for Claude Code, Opus, or Research. That fits students, casual users, and people trying Claude before paying.

Claude Pro: Power User Plan

Cost: $20/month billed monthly, or $17/month on the annual plan ($200 billed up front) (source)

Plan features (plan comparison unless noted):

• At least 5x the Free plan’s usage per 5-hour session, with weekly limits
• Opus, Sonnet, and Haiku models. Opus 5 is the strongest model included in Pro (source)
• Fable 5 and Fable 5.1 through pay-as-you-go usage credits, outside the plan’s limits (source)
• Claude Code
• Projects (Free is capped at 5) and the Research tool
• Claude Design, Slides, and Docs, Claude Science, and Claude in Chrome and Microsoft 365
• Claude Cowork (source)
• Memory, Artifacts, and custom connectors, so Claude can read your own tools. ScreenApp has one: see connecting ScreenApp to Claude with MCP

Priority access at high traffic times is listed for Max, not Pro, on the plan comparison.

Cowork update: On 16 September 2026 Anthropic announced that Cowork and chat are merging into one Claude, rolling out to Pro and Max first, with Team and Free to follow (announcement).

Best for (editorial judgement): Criteria: regular weekday use, a need for Opus or Claude Code, and one person per account. Pro fits most individual professionals on those criteria.

Claude Max: Premium Tiers

Max is billed monthly only and comes in two usage tiers (What is the Max plan?).

Claude Max 5x ($100/month)

Cost: $100/month (source)

Features (plan comparison unless noted):

• 5x Pro’s usage per 5-hour session
• Everything in Pro, plus higher output limits and early access to advanced features
• Priority access at high traffic times
• Opus 5 as the default model (source)
• Fable 5 and 5.1 included, using up to 50% of your weekly limits at no extra cost (source)

Best for (editorial judgement): Criteria: you hit Pro’s session or weekly limits on most working days, or you want Fable inside your plan instead of paying usage credits.

Claude Max 20x ($200/month)

Cost: $200/month (source)

Features:

• 20x Pro’s usage per 5-hour session (source)
• Everything in Max 5x, including priority access and Fable within 50% of weekly limits (source)

Best for (editorial judgement): Criteria: all-day use, long Claude Code sessions, or regular heavy Fable work where Max 5x still runs out. Accounts are for one person; teams should look at Team Premium seats instead.

Team and Enterprise Plans

Claude Team Plan

Cost: Standard seat $20/month billed annually or $25 monthly. Premium seat $100/month billed annually or $125 monthly. For teams of 2 to 150 (source).

Features:

• Standard seats get more usage than Pro, and Premium seats get 5x more than Standard (source). Anthropic’s help center puts these at 1.25x and 6.25x Pro (source)
• Claude Code and Cowork, Claude Design, Slides, and Docs, and Claude Science
• Microsoft 365, @Claude, and enterprise search across your organization
• Single sign-on, central billing and administration, usage analytics, and project sharing
• Mix and match seat types
• Fable on Premium seats within 50% of weekly limits; Standard seats use usage credits (source)

Feature list from the Team plan comparison unless noted.

Best for (editorial judgement): Criteria: two or more people, a need for admin controls, SSO, or shared projects, and central billing. On annual billing a Standard seat costs $20, the same as Pro’s monthly price, so teams that want admin features pay no seat premium over individual monthly Pro accounts.

Claude Enterprise Plan

Cost: $20/seat/month billed annually, plus usage billed at API rates, so cost scales with the models and tasks your team runs. Available self-serve or through sales (source).

Features (plan comparison):

• Everything in Team
• Admin-set user and organization spend limits
• Role-based access, SCIM, audit logs, and a Compliance API
• Custom data retention controls, network-level access control, and IP allowlisting
• HIPAA-ready offering available
• Claude Security (beta)

Best for (editorial judgement): Criteria: identity management (SCIM), audit and compliance requirements, or a need for regulated-data options such as the HIPAA-ready offering.

API Pricing Guide

Current Claude API Costs

Claude’s API pricing is per million tokens (MTok), charged separately for input (your prompts) and output (Claude’s responses). The Opus tier has stayed at $5/$25 from Opus 4.5 through Opus 5 (source).

Standard API pricing (per million tokens) (API pricing):

Sonnet 5’s $2/$10 price was announced as introductory pricing through August 31, 2026. Anthropic has since made it the standard price and cancelled the planned rise to $3/$15 (source). Legacy Sonnet 4.6 and 4.5 still cost $3/$15.

Tokens and words: Claude Opus 4.7 and later models use a newer tokenizer that produces about 30% more tokens for the same text (source). Anthropic puts 1M tokens at roughly 555,000 English words on the current tokenizer, compared with about 750,000 on older models (source). Count tokens with Anthropic’s token counting endpoint before you budget.

Fast Mode

Fast mode is a research preview that gives up to 2.5x higher output tokens per second on Opus 5 and Opus 4.8, with the same model and no change to intelligence. It costs $10/$50 per million tokens across the full context window, needs a beta header, is available on the Claude API only, and does not work with the Batch API (fast mode docs). Access is by request or waitlist.

Cost Optimization Features

Batch API (50% Discount)

The Batch API processes requests asynchronously at 50% off input and output tokens (source):

• Haiku 4.5 Batch: $0.50/$2.50 (vs $1/$5 standard)
• Sonnet 5 Batch: $1/$5 (vs $2/$10 standard)
• Opus 5 Batch: $2.50/$12.50 (vs $5/$25 standard)
• Fable 5.1 Batch: $5/$25 (vs $10/$50 standard)

Suits (editorial judgement): bulk document processing, overnight analysis, and content pipelines where you do not need an instant answer.

Prompt Caching

Prompt caching cuts the cost of re-sending the same large context (codebases, legal documents, style guides). Multipliers are relative to the base input price (source):

• 5-minute cache write: 1.25x base input
• 1-hour cache write: 2x base input
• Cache read: 0.1x base input on most models, and 0.025x on Fable 5.1

Caching costs for Sonnet 5 (per million input tokens): $2.50 for a 5-minute write, $4 for a 1-hour write, $0.20 for a cache read (source).

Worked example (arithmetic on published prices, not a measured bill): sending the same 150K-token codebase to Sonnet 5 20 times within the cache window.

• Without caching: 20 × (150K × $2/M) = $6.00 input cost
• With caching: (150K × $2.50/M) + 19 × (150K × $0.20/M) = $0.375 + $0.57 = about $0.95
• Difference: about $5.05, an 84% reduction on input cost

Long Context Pricing

Claude 4.6 and later models include the full 1M token context window at standard pricing, so a 900K-token request is billed at the same per-token rate as a 9K-token request (source). There is no separate long-context premium on the current models.

API Tool and Region Pricing

Server-side tools and data residency have their own charges (API pricing):

• Web search: $10 per 1,000 searches, plus standard token costs for search results
• Web fetch: no charge beyond standard token costs
• Code execution: 1,550 free hours per organization per month, then $0.05 per hour per container. It is free when used with the current web search or web fetch tools
• US-only inference: 1.1x multiplier on all token pricing for Claude 4.6 and later models

API vs Subscription Analysis

Anthropic does not publish subscription limits in tokens, so there is no exact break-even point between Pro and the API. What you can do is compare what $20 buys on each side.

On the API (arithmetic on published prices): $20 pays for about 5 million input tokens plus 1 million output tokens on Sonnet 5 ($10 + $10), or about 2 million input plus 400,000 output tokens on Opus 5 ($10 + $10) (prices). Long documents and agent runs use tokens quickly, especially on the newer tokenizer.

On Pro: $20/month buys a usage allowance that resets every five hours, with weekly caps, plus Claude Code, Projects, Research, and the chat apps (source).

Editorial judgement: Criteria: how steady your usage is and whether you need the chat apps. Light, irregular, or automated usage tends to cost less on the API because you pay only for what you send. Daily interactive use tends to cost less on a subscription, and you get the apps and Claude Code with it. Track a week of token use in the Claude Console before deciding.

For Developers:

For developers, the API vs. subscription decision depends heavily on real usage patterns. Sporadic API integrations, automation tasks, or batch processing jobs typically favor API pricing because costs align directly with consumption. However, teams building solutions that require continuous model access often find that Pro or Max subscriptions remove budget unpredictability while enabling faster iterations and uninterrupted development cycles.

Claude vs ChatGPT Pricing (2026)

We could not verify OpenAI’s current prices against OpenAI’s own pages on 18 September 2026, so this section lists the Claude side only. Check ChatGPT pricing and OpenAI API pricing directly for the other half of the comparison.

How to compare (editorial judgement): Criteria: context length you need, the media types you work with, and API cost at your volume. Claude’s 1M context on current models suits long documents, large codebases, AI document analysis, and legal and contract review. Check each vendor’s plan page for the media features you need. For API work, compare per-token prices at your own input and output mix, since the cheaper option depends on how output-heavy your workload is.

Is Claude Open Source?

No, Claude AI is not open source. Anthropic keeps Claude’s model weights, architecture, and training data proprietary. Unlike open-weight alternatives (Llama, Mistral, DeepSeek), you cannot download and self-host Claude models.

Anthropic does publish research papers on its Constitutional AI method, model system cards, and safety testing results.

Open-weight alternatives to consider:

• Llama (Meta): free to download and self-host under Meta’s license
• Mistral (Mistral AI): European models, several with open weights
• DeepSeek (DeepSeek): open-weight reasoning models, popular for cost-sensitive deployments

Trade-offs: Self-hosted open models remove subscription costs and keep data on your own infrastructure, but they need technical expertise to deploy and run.

Choosing Your Claude Plan (2026 Recommendations)

Everything in this section is editorial judgement based on published prices and plan features, not on our own testing.

For students and casual users: Criteria: occasional use, no need for Opus or Claude Code. Start with Free and upgrade to Pro only if you hit limits several days a week.

For developers: Criteria: daily interactive coding vs automated or irregular calls. Daily Claude Code users fit Pro, then Max if limits bite. Integrations and batch jobs fit the API, where the Batch API halves token costs (source).

For small teams (2 or more people): Criteria: need for admin controls, SSO, or shared projects. Team Standard seats cost $20/month on annual billing and Premium seats $100/month (source).

For larger teams: Criteria: coordination across departments, usage analytics, and enterprise search. Team covers up to 150 seats with central billing and admin (source), which helps when coordinating collaborative meeting tools and AI workflows.

For enterprises: Criteria: SCIM, audit logs, compliance tooling, or custom data retention. Enterprise is $20/seat/month billed annually plus usage at API rates (source).

The bottom line (editorial judgement): Criteria: usage volume, need for Claude Code or Opus, and team admin needs. Most individual professionals start with Pro, casual users stay on Free, heavy users move to Max, and developers should compare API costs against a subscription using their own token counts.

---

Sources (checked 18 September 2026):

• Claude pricing (plans and plan comparison)
• Claude API pricing documentation
• Claude models overview
• Fast mode documentation
• What’s new in Claude Fable 5.1
• Introducing Claude Opus 5
• Introducing Claude Sonnet 5
• What is the Max plan?
• What is the Team plan?
• Claude Fable models on your plan
• Get started with Claude Cowork
• Claude Cowork and chat are now one Claude

How we put this together: this is a pricing guide based on Anthropic’s published documentation, pricing pages, help center articles, and announcements, checked on 18 September 2026. ScreenApp did not test the plans, models, or features described.