Claude AI FAQ — Quick Answers about Models, Pricing, Apps & API

This Claude AI FAQ gives direct answers about Claude models, subscriptions, API costs, apps, safety, and common support issues.

c-ai.chat is the independent guide to Claude AI; Anthropic makes Claude, and claude.ai is the official chat product.

For narrower help, use our Claude pricing guide, Claude model guide, Claude API guide, Claude features guide, and Claude resources.

• Independent · c-ai.chat is not Anthropic.
• Checked against official pages · plans, models, API docs, support, trust, and status.
• Two pricing paths · subscriptions for Claude apps; per-token billing for the API.

Table of contents

• The basics
• Plans & pricing
• API pricing
• Models
• Apps & platforms
• Claude Code
• Safety & data
• Comparisons
• Troubleshooting & status
• Sources

The basics

Start here if you are new to Claude.

What is Claude AI?

Claude AI is Anthropic’s family of AI assistants and models. You can use Claude in the official chat app to draft text, analyze documents, write code, compare ideas, and answer questions. Developers can call Claude models through the API. Claude can still be wrong, miss context, or refuse unsafe requests, so review important work before you rely on it.

Who makes Claude AI?

Claude is made by Anthropic, an AI company focused on large language models and AI safety. Anthropic runs the official Claude product, developer platform, support pages, trust center, and status page. c-ai.chat explains that ecosystem independently and does not speak for Anthropic.

Do I need an account to use Claude?

Yes. Claude requires a free account—there is no anonymous access. When you create an account, phone verification is mandatory for new users and cannot be skipped, and the minimum age to use Claude is 18. You can sign up at claude.ai using an email magic link, Google, Apple, or, on eligible plans, enterprise SSO.

Is Claude available in my country?

Availability depends on region, product, account type, and local rules. The fastest check is the sign-up flow at claude.ai. If you cannot create an account, add a billing method, or access an app, check Claude Support.

Is c-ai.chat the official Claude site?

No. c-ai.chat is an independent guide. We are not Anthropic, and we do not impersonate claude.ai. Use c-ai.chat for explanations, comparisons, and practical context. Use claude.ai for the official Claude app and platform.claude.com for the official API platform.

Why is it called Claude?

The name is often speculated to nod to Claude Shannon, whose work shaped information theory, but Anthropic has not formally confirmed this. The practical distinction matters more: Claude is the assistant and model family; Anthropic is the company; claude.ai is the official chat product.

Plans & pricing

Claude subscriptions cover the app and workspace products. API usage is billed separately.

Is Claude AI free?

Yes. Claude has a Free plan at $0/month. It gives access to the web, mobile, and desktop apps with session-based usage limits plus separate weekly limits, and it includes Claude Sonnet 5 and Claude Haiku 4.5. Free is enough for casual testing, short chats, and occasional writing help. If you hit limits often or need paid features, compare the paid plans on the official Claude pricing page.

How do Claude’s usage limits work?

Claude’s app plans use session-based limits with periodic resets rather than one fixed message quota. A usage session resets roughly every five hours, and there are separate weekly limits on top of that; on higher tiers, Opus-class usage can have its own weekly reset apart from other models. How much you can do in a session varies with the model, message length, attached files, and the tools or features you use. Paid plans (Pro, Max, Team, and Enterprise) raise these limits. See Anthropic’s usage-limit best practices for current specifics.

What does Claude Pro include?

Claude Pro costs $20/month, or $17/month with annual billing (billed at $200 up front). It is the main paid plan for individuals. Pro gives more usage than Free and adds the Opus 4.8 and Fable 5 models, plus Claude Code, Cowork, and Projects.

What is Claude Max for?

Claude Max is for users who need much more capacity than Pro. It starts at $100/month and offers 5x or 20x Pro usage. It fits long coding sessions, large document work, heavy research, and frequent high-output tasks, with higher output limits, early access, and priority traffic where Anthropic lists it.

What is the difference between Team Standard and Team Premium?

Team Standard is $25/seat/month, or $20/seat/month with annual billing. It is the lower-cost team plan, with a shared workspace, SSO, and admin controls. Team Premium is $125/seat/month, or $100/seat/month with annual billing—the higher-capacity team tier with priority traffic and expanded admin controls. Choose based on capacity, governance, and admin needs.

Does Enterprise have a fixed price?

Enterprise is not a simple self-serve subscription. Anthropic lists Enterprise as a $20/seat base plus usage that scales at API rates, with custom contracts. Enterprise controls can include SCIM, audit logs, role-based access, spend controls, and regional data residency. Final terms depend on the organization and contract.

Can I switch plans at any time?

Plan changes are handled through Claude account or workspace billing controls. Individual upgrades are usually self-serve. Team and Enterprise changes may require an admin or Anthropic contact. Check the billing screen before changing plans, because seat counts, usage limits, renewal terms, and workspace ownership can affect the result.

How do I cancel?

Use the billing or plan settings in the official Claude account interface. If you are part of a Team or Enterprise workspace, ask the workspace admin first. If the cancellation option is unclear, use Claude Support. Do not cancel through c-ai.chat; we do not manage Claude accounts or billing.

API pricing

The Claude API is billed by input tokens and output tokens. Prices below are USD per million tokens.

Model	Role	Input price	Output price	Context and output notes	Best use
Claude Fable 5	Most capable (widely released)	$10/M tokens	$50/M tokens	1M-token context; 128K max output	Highest-capability reasoning and generation
Claude Opus 4.8	Complex agentic coding & enterprise	$5/M tokens	$25/M tokens	1M-token context; 128K max output	Hard agentic coding, difficult analysis, enterprise work
Claude Sonnet 5	Default model (all plans)	$2/M tokens (introductory, through 2026-08-31; then $3/M)	$10/M tokens (introductory, through 2026-08-31; then $15/M)	1M-token context; 128K max output	Balanced writing, coding, analysis, and agents
Claude Haiku 4.5	Fast / low cost	$1/M tokens	$5/M tokens	200K-token context; 64K max output	High-volume, latency-sensitive, simpler tasks

How is the Claude API priced?

The Claude API is priced per million input tokens and per million output tokens. Input tokens include text, tool results, files, and context you send to the model; output tokens are what the model generates. Current rates: Fable 5 is $10/M input and $50/M output; Opus 4.8 is $5/M input and $25/M output; Sonnet 5 is $2/M input and $10/M output as introductory pricing through 2026-08-31, after which it moves to $3/M input and $15/M output; and Haiku 4.5 is $1/M input and $5/M output. Legacy models such as Opus 4.7 ($5/M and $25/M) and Sonnet 4.6 ($3/M and $15/M) remain available but are superseded. Confirm billing details in the official API pricing docs.

What is prompt caching?

Prompt caching reduces the cost of repeated input. If your app sends the same long instructions, policy text, codebase summary, or document prefix many times, Anthropic can treat the reusable part as cached input. Cached input tokens receive 90% off. Caching is most useful when many requests share the same large context.

What is the Batch API discount?

The Batch API is for asynchronous jobs that do not need an immediate response, such as classification runs, offline document processing, or bulk analysis. Batch API usage receives 50% off both directions. Use it when latency is flexible. Do not use it for interactive chat or live user workflows.

Can I get the 1M-token context window without an Enterprise contract?

Yes. Fable 5, Opus 4.8, and Sonnet 5 support a 1M-token context window at standard API rates (Haiku 4.5 supports 200K tokens). You still need access to the relevant model and enough rate limit for your workload. Large context helps with codebases, long contracts, multi-file research, and document comparison, but it does not replace clear prompt structure.

How do I estimate API cost in advance?

Estimate tokens first, then apply the model’s input and output rates. Keep output limits tight. Cache repeated context. Use Batch API when work can run asynchronously.

1. Choose the model

   Use Opus 4.8 (or Fable 5 for the most capable) for the hardest tasks, Sonnet 5 as the default, and Haiku 4.5 for low-cost speed.

2. Estimate input and output tokens

   Count system prompts, user messages, retrieved documents, tool results, and expected responses. Long files can dominate cost.

3. Apply the formula

   estimated_cost = (input_tokens / 1,000,000 * input_rate) + (output_tokens / 1,000,000 * output_rate)

4. Reduce repeated input

   Use prompt caching for stable context and Batch API for offline jobs when the workflow allows it.

What is the cheapest Claude model?

Claude Haiku 4.5 is the lowest-cost current model at $1/M input and $5/M output. Start with Haiku for short extraction, routing, tagging, lightweight support automation, and high-volume workflows. If quality drops on complex tasks, move to Sonnet 5 before using Opus 4.8.

Models

Model choice affects quality, latency, context size, output length, and cost.

Need	Best first model	Why
Highest capability on the hardest work	Claude Fable 5	Most capable widely released Claude model; 1M-token context; 128K max output
Complex agentic coding and enterprise tasks	Claude Opus 4.8	Flagship agentic model; 1M-token context; 128K max output
Everyday writing, coding, analysis, and agents	Claude Sonnet 5	Default on all plans; strong balance of quality, speed, and cost; 1M-token context
High-volume, latency-sensitive, simpler tasks	Claude Haiku 4.5	Fastest, lowest-cost current model

Which Claude model should I use?

Use Sonnet 5 for most work. It is the default across all plans and fits writing, coding, analysis, customer operations, and agent workflows. Step up to Opus 4.8 for complex agentic coding, difficult analysis, and enterprise tasks, or to Fable 5 when you need the most capable model available. Use Haiku 4.5 when cost and speed matter most. Opus 4.7 and Sonnet 4.6 are legacy models—still available, but superseded by Opus 4.8 and Sonnet 5. Check the official model overview before locking production code to a model ID.

What is Claude Fable 5?

Fable 5 is Anthropic’s most capable widely released model. It suits the hardest reasoning and highest-quality generation, with a 1M-token context window and up to 128K output tokens. On the API it is $10/M input and $50/M output. In the apps it is available on Pro, Max, Team, and Enterprise plans.

What is Opus 4.8 best for?

Opus 4.8 is Anthropic’s flagship agentic model, recommended for complex agentic coding, difficult analysis, and enterprise workloads. It supports a 1M-token context window and up to 128K output tokens, and on the API it is $5/M input and $25/M output. Use it when task quality justifies the cost; for the very highest capability, consider Fable 5.

What is Claude Mythos 5?

Mythos 5 is a limited-availability model offered by invitation only, through Anthropic’s Project Glasswing for approved customers, and aimed at defensive-cybersecurity workflows. It is not self-serve and shares Fable 5’s specifications and pricing. If you have not been invited, use the generally available models above.

Can I still use legacy models like Opus 4.7 or Sonnet 4.6?

Opus 4.7 and Sonnet 4.6 are legacy models. They remain available and priced (Opus 4.7 at $5/M input and $25/M output; Sonnet 4.6 at $3/M input and $15/M output), but they are superseded by Opus 4.8 and Sonnet 5 and should not be treated as current, latest, or default. Older generations may also still be available to some accounts. For new builds, start with current models, and if you depend on a legacy model, plan a migration path so your app does not rely on a model that may be retired later.

What is a tokenizer, and why should I retest counts?

A tokenizer splits text into tokens before a model processes it. Tokens are the billing and context units used by the API. Different tokenizers can count the same text differently. If you move prompts to a new model, test token counts rather than assuming old estimates, truncation rules, and prompt budgets still hold.

Does the model affect chat or only the API?

It can affect both. In the Claude app, available models depend on plan, usage limits, and product settings. In the API, developers select a model ID directly. The same model family can feel different in chat and API because the app adds interface features, file handling, projects, tools, and product-level limits.

Apps & platforms

Claude is available through the official app experience and Anthropic’s developer platform.

Where can I use Claude?

You can use Claude in the official web app at claude.ai, on supported mobile apps, on desktop apps, and through Anthropic’s API platform. Businesses can also use Claude through Team, Enterprise, and supported cloud platforms. Access can vary by region, plan, and workspace policy.

Is Claude available on iOS and Android?

Yes. Claude is available on iOS and Android as part of the official Claude app experience. Use official Anthropic links from claude.ai or your device’s app store listing. Avoid unofficial apps that copy the Claude name, ask for credentials, or claim to be Anthropic without proof.

Are there desktop apps for Mac, Windows, and Linux?

Yes. Mac and Windows desktop apps are generally available. Anthropic also offers an official Linux desktop app in beta for Ubuntu and Debian (x64 and arm64), installable from an apt repository or a .deb package; it supports Chat, Cowork, and Claude Code, though some features such as Computer Use and dictation are not available yet, and Fedora and RHEL are not yet supported. Download only from the official page at claude.com/download so you do not install a third-party wrapper by mistake.

Can I use Claude in Excel, Word, or PowerPoint?

Yes, on paid plans. Claude for Microsoft 365 works with Excel, Word, and PowerPoint—including native slide generation with charts and diagrams—and Outlook support is in beta. Use these for spreadsheet analysis, document drafting, slide creation, and email work, and review outputs before sending business-critical material.

Does Claude work on AWS Bedrock or Google Vertex AI?

Yes. Claude models are available through selected cloud platforms—including Amazon Bedrock, Google Cloud (Vertex AI), and Microsoft Foundry—as well as Anthropic’s direct API. Cloud-platform access can differ from the direct API in model IDs, regions, quotas, authentication, logging, and billing. If you already run workloads on a cloud platform, compare that path with Anthropic’s API platform before choosing.

Claude Code

Claude Code is Anthropic’s official coding agent.

What is Claude Code?

Claude Code helps developers inspect repositories, edit files, explain code, run approved commands, and work through implementation tasks from the command line or on the web. It is built for real coding workflows, so permissions, review, tests, and source control still matter. Start with the official Claude Code docs for setup.

Does Claude Code need Pro or Max?

Claude Code is available on Pro, Max, Team, and Enterprise plans (and through the developer console), but not on the Free plan. It is subject to your plan’s usage limits, and Max is aimed at heavier users who need more capacity. Team and Enterprise access can depend on workspace settings and admin policy. If you plan to use Claude Code all day, compare expected usage with Pro and Max before choosing a plan.

What is the Claude Agent SDK?

The Claude Agent SDK is a developer toolkit for building agent workflows around Claude. It helps define instructions, tools, context, permissions, and execution patterns for multi-step tasks. Use the SDK when a simple one-shot API call is not enough and you need repeatable agent behavior.

Can I run Claude Code in CI?

Yes, if your account, authentication method, and repository controls support it. Treat CI usage carefully. Use scoped credentials, restrict write access, log outputs, run tests, and keep human review on protected branches. Do not give an agent unrestricted deploy permissions unless your organization has approved that risk.

What are managed agents?

Managed agents are organization-controlled agent setups rather than one-off local prompts. They can package instructions, tools, permissions, and workflow expectations so teams can reuse approved agent behavior. Exact controls depend on Anthropic’s current Claude Code and workspace documentation, so confirm details in the official docs before rollout.

Safety & data

Privacy, security, and compliance depend on product path and contract terms.

Does Anthropic train on my data?

Anthropic’s official privacy policy and product terms control this answer, and data handling differs across consumer chat, API, Team, and Enterprise. Retention and training are separate concepts; if you opt into training on consumer chat, de-identified data may be retained for up to five years, while incognito chats are not used for training and are not saved to your history (though they may be retained on servers for a limited period). Review the Claude privacy center and your plan terms before sending sensitive data.

Where is my data stored?

Storage depends on product, account type, and contract. Enterprise includes regional data residency options. Self-serve users should not assume a specific storage region unless Anthropic states it for their plan. For compliance reviews, use the Anthropic Trust Center and your contract documents rather than third-party summaries.

Is Claude suitable for HIPAA-regulated data?

Requirements for protected health information depend on your plan and contract, not on the app defaults. If you handle PHI, confirm current compliance options, any required agreements (such as a Business Associate Agreement), access controls, retention settings, and approved use cases directly with Anthropic before sending regulated health data to Claude. Do not assume a personal plan is appropriate for PHI.

Where can I find Claude’s security and compliance information?

Anthropic publishes security and compliance materials through its Trust Center. If your procurement or security team needs specific attestations, request them through the official portal at trust.anthropic.com rather than relying on screenshots, old questionnaires, or unofficial claims.

Where do I report a security issue?

Report security issues through Anthropic’s official security or trust channels, starting at trust.anthropic.com. Do not post tokens, prompts, private files, or exploit details publicly. If the issue affects your account rather than Anthropic’s systems, contact Claude Support.

Can I delete my chat history?

Claude provides account and chat controls in the official product. You can delete individual chats or manage account data through app settings where available. Team and Enterprise workspaces may have admin retention rules. If you cannot find the control you need, use Claude Support rather than sending account details to an unofficial site.

Comparisons

Compare AI assistants on your actual work, not brand preference.

Assistant	Maker	Common strength	Check before choosing
Claude	Anthropic	Long-context reasoning, writing, coding, and agent workflows	Plan limits, model access, and API cost
ChatGPT	OpenAI	Broad consumer ecosystem and many integrations	Model choice, tool access, data settings, and pricing
Gemini	Google	Google ecosystem alignment and multimodal workflows	Workspace fit, model behavior, and cloud deployment path

How does Claude differ from ChatGPT?

Claude and ChatGPT are general AI assistants from different companies. They use different models, tools, interfaces, and policies. Claude is often chosen for long-context reading, writing quality, coding assistance, and Anthropic’s safety approach. ChatGPT may fit users who prefer its ecosystem or specific integrations. Test both on your real tasks.

How does Claude differ from Google Gemini?

Claude is Anthropic’s assistant and model family. Gemini is Google’s assistant and model family. The best fit depends on your workspace, cloud stack, data controls, model behavior, and pricing. Claude may fit long document analysis and coding workflows. Gemini may fit teams already centered on Google products.

Is Claude better at coding than alternatives?

Claude is strong at coding, especially with Sonnet 5, Opus 4.8, and Claude Code. “Better” depends on the repository, framework, tests, task clarity, and review process. A practical benchmark is simple: give each assistant the same issue, same files, same tests, and same time budget, then compare passing tests and code review quality.

Does Claude generate images?

Claude can analyze images when the selected model and product path support vision, but Claude is not primarily an image-generation product. It can write prompts, create SVG or HTML diagrams, describe visual ideas, and help plan graphics. If your main job is producing image files, use a dedicated image model or design tool.

When should I not use Claude?

Do not use Claude as the sole authority for legal, medical, financial, safety-critical, or regulated decisions. Do not use it with confidential data unless your plan and contract allow it. Do not use it where you need guaranteed deterministic output, verified live facts without sources, or native image generation. Human review still matters.

Troubleshooting & status

Use official status and support channels for account, billing, outage, and access issues.

Where do I check if Claude is down?

Check status.claude.com. That is Anthropic’s official status page for Claude services. If the status page shows an incident, wait for Anthropic’s updates. If the status page looks normal, the problem may be local network access, account limits, browser state, workspace policy, or API rate limits.

Why am I seeing rate-limit errors?

Rate-limit errors mean your account, workspace, or API key is sending more work than the allowed limit. Limits can involve requests, tokens, concurrency, or sudden traffic increases. Reduce parallel calls, add retry logic with backoff, shorten prompts, cap outputs, and spread large jobs over time. For API work, review Anthropic’s platform docs and your account limits.

Why won’t Claude answer a question?

Claude may refuse if the request appears unsafe, asks for wrongdoing, requests private data, or conflicts with Anthropic’s policies. It may also stall if the prompt is vague, overloaded, or missing required context. Rephrase the request, narrow the task, explain the legitimate goal, and remove requests for harmful or private information.

How do I contact Anthropic support?

Use support.claude.com. That is the official support site for account, billing, product, and access issues. c-ai.chat cannot view your Claude account, cancel a subscription, restore access, change limits, or investigate billing. Never send account secrets to an independent guide.

How do I reset my password?

Use the official Claude sign-in flow and password reset option. If you sign in with Google, Apple, or SSO, you may need to reset credentials through that identity provider instead. If you are in a Team or Enterprise workspace, your admin may control access. Use Claude Support if the reset flow fails.

Independent guide. Not affiliated with Anthropic. For the official Claude product, visit claude.ai.

Facts verified: 2026-07-13