Claude 5: What you need to know about Anthropic's AI models and chatbot

I've been using Claude long enough to remember when the main selling point was that it was a nicer chatbot to talk to than the alternatives. (That's still true, for what it's worth.) But Claude no longer just talks to you about your work; it also does your work for you. You can give Claude a project, head off to make a coffee, and check in occasionally when questions pop up.

For enterprises looking to get real productivity gains from AI, Claude has become the default choice. And Claude is equally popular at smaller companies among developers, writers, and teams who just want an AI that can handle complex work.

In this article, I'll explore what Claude can do and why it's become the default AI for enterprises and professionals.

Table of contents:

• What is Claude?

• What can Claude do?

• How to try Claude for yourself

• Automate Claude with Zapier

What is Claude?

Claude is a family of large language models (LLMs) developed by Anthropic, a company founded in 2021 by former OpenAI researchers. Anthropic centers its development of Claude around a constitution that guides Claude during the training process to prioritize safe, ethical, and helpful AI interactions.

Claude is also the name of the AI chatbot built on top of these models. It's a general-purpose assistant that can handle tasks like writing, analysis, coding, and reasoning.

Claude's model family: Fable, Opus, Sonnet, Haiku

Anthropic offers four main model variants, each with different performance characteristics and price points.

• Claude Fable 5 is Anthropic's most capable publicly available model. It's the most advanced model for long autonomous coding projects, deep research, and complex data analysis. For regular knowledge work, it's often overkill (it consumes 2x the tokens of Opus). Fable is the public version of Claude Mythos, a model that Anthropic withheld from public release after determining it could pose a cybersecurity threat; to make Fable safer, Anthropic added built-in guardrails that route some queries to Opus instead.

• Claude Opus 4.8 is the best model for most complex tasks, and the right choice for any heavy work that doesn't need the frontier-model firepower of Fable. It's well-suited to research, agentic coding, and enterprise applications that require high levels of accuracy. At half the cost of Fable, it offers significant cost savings while delivering similar performance for many use cases.

• Claude Sonnet 4.6 is the best model for all-around performance. Fable and Opus still have an edge on the most complex tasks, but at $3 per million input tokens and $15 per million output tokens, Sonnet is the best default model for most users due to its balance of performance and cost.

• Claude Haiku 4.5 is optimized for speed and affordability. At $1 per million input tokens and $5 per million output tokens, it's a good choice for time-sensitive use cases like customer service or for affordably processing lots of data.

Anthropic has also released its Mythos 5 model to a small group of cybersecurity and infrastructure firms through Project Glasswing, an initiative that allows trusted organizations to access Anthropic's most powerful models.

Claude API pricing

How to access Claude

If you're using Claude's free plan, you get limited usage of Sonnet and Haiku. Claude Pro, at $20/month, gives you access to Opus and Fable along with features like Claude Code, Claude Cowork, and Research. If you're a power user, Claude's $100/month Max 5x plan increases your usage limit by another 5x (you can also pay $200/month to increase your limit by 20x). For current details on Fable 5 availability across plans, check Claude's pricing page.

You can also access all of these state-of-the-art models on Zapier, bringing the power of Claude's AI into all the apps you already use. And with Zapier MCP, you can give Claude safe, governed access to 9,000+ apps, directly from the chat window.

What can Claude do?

Claude is best for use cases like coding, deep reasoning, nuanced content creation, and privacy-focused enterprise applications. Here are the features that help Claude stand out.

Artifacts

Claude Artifacts lets you build interactive apps, tools, and content. They're generated automatically when Claude thinks you might want to edit, share, reuse, or reference the content later. Artifacts open up in a split screen, which makes it easy to continue iterating by conversing with Claude, and they can be used to create things like data visualizations, small web apps, games, interactive React components, code snippets, or documents.

You might be surprised how useful this is for satisfying your everyday curiosity. Seriously: ask Claude to generate data visualizations for random questions and see what happens. It still blows my mind that from the comfort of my couch, I can ask Claude for "an interactive visualization showing which countries export the most bananas per capita" and get a masterpiece like the chart below, complete with one-click filters and expandable detail panels.

Artifacts are a productivity tool too. You can spin up internal business apps in minutes, then share them with your team with one click. And since Anthropic now allows you to create AI-powered artifacts that consume each user's tokens—rather than yours—it's easier than ever to incorporate AI features into the apps, dashboards, and reports you build with Claude.

In the example below, I used Fable to create a shareable project tracker that uses AI to generate tasks and prioritize next steps.

Research

All of the leading LLMs have some form of "deep thinking" mode. Claude offers two variations: research and extended thinking mode.

Like ChatGPT's Deep Research, Claude's Research feature looks through dozens (and often hundreds) of sources to create comprehensive reports addressing nuanced requests. For example, I asked Claude to do a comprehensive review of Alaska Inner Passage cruise options, specified a few parameters, and got a thoroughly-researched recommendation in ten minutes.

What's especially neat is that Claude automatically takes a multi-agent research approach to speed things up. It starts by mapping out a research plan, then delegates each task to a different subagent so that Claude can investigate each thread of research simultaneously.

Before running a research task, it's worth knowing about Claude's effort levels, which apply to all tasks but are particularly relevant to research. In addition to choosing your model, you can set the effort level to Low, Medium, High, or Max. Higher effort means more thorough responses, but it takes longer to complete tasks and you'll burn through your usage limits faster. For serious research tasks, High or Max effort levels give you the best results. For quicker questions, Low effort is a better default.

If you just need Claude to reason its way through a problem and you don't need to check web sources, extended thinking mode is a good option. It employs a "hybrid reasoning" approach that lets the AI either respond almost instantly or take time to reason step-by-step before answering. Either way, you'll be able to see how Claude arrived at its conclusion, which builds confidence in the answers it gives you and prevents unnecessary back-and-forth.

When your question can be accurately handled by quickly pulling from its training data, Claude will tell you.

And when you ask a question that requires analysis, probabilities, and complex ethical or societal considerations, Claude will ponder your question a bit longer and explain all the factors that are driving its thought process.

Projects and Skills

If you use Claude regularly for the same kind of work, Projects lets you create a persistent workspace with saved instructions, files, and context, so you don't have to reexplain yourself every time you start a new task.

For example, let's say you're a product manager who reviews customer feedback every day. Normally, you'd need to introduce your product, roadmap, and preferred response format during every single session. Projects lets you anchor all of your chat activity with the same instructions and long-term memory context, while also uploading materials to a persistent file archive so Claude always has the context it needs.

Projects is great for consistently referencing the same context and background knowledge. But if you're looking to define a repeatable process, that's when you'll want to reach for Claude Skills.

A Skill is a reusable set of instructions that outlines the process and output format Claude should adhere to. Want Claude to always draft newsletters in the same way, or follow a specific editing checklist when you share a draft? That's a Skill. You can build your own by simply asking Claude to help you build a Skill, which will trigger—you guessed it—a skill-creator Skill to walk you through the process.

Skills and Projects work best together. Use Projects as a container to hold the context of what you're working on, and use Skills to help you take action in a predictable, repeatable way.

Claude Code

Claude Code, Anthropic's agentic coding tool, is wildly popular for coding. Among enterprises, it owns 54% of the coding market. And among developers, a February 2026 survey found that 71% of those who use AI coding agents use Claude Code.

Rather than simply generating code, Claude Code can execute entire projects from start to finish. Just give Claude your vision ("build me an iOS app that tracks blimps"), and it'll take all the necessary steps to plan and execute it, while checking in occasionally for questions along the way.

You get full control over how often you want Claude Code to check in while it runs commands and modifies files. By default, it asks for your approval at each checkpoint, but you can also select Auto mode if you only want to be in the loop if a possible safety issue arises. There's also an ominously-named --dangerously-skip-permissions option, which bypasses all safety prompts so Claude Code can run uninterrupted. (Ideally, you'd only use this in sandboxed environments.)

Claude Code's hours-long autonomous projects are made possible through a couple of innovations. Compaction allows Claude to summarize its own progress so it doesn't hit a context wall, and agent teams coordinate multiple Claude instances that work in parallel on parts of a larger task. And since Claude now has a 1M token context window, it can hold entire codebases, meaning it can understand the entire context of your project without issue.

With frontier models like Fable, multi-hour agentic coding projects can be tackled more independently than ever before. Ethan Mollick, an AI specialist and professor at the Wharton School of the University of Pennsylvania, likens Fable's process to a patron commissioning work from an artist: you can define what you're looking for and approve or reject the outcome, but you don't have much influence over the dozens of micro-decisions that happen in between.

While Fable can be overkill for smaller projects, one area where it might be worth the extra tokens is design: it has better taste than other Claude models. In my testing, Fable was more likely than other models to one-shot a landing page that gets everything right on the first go.

Developers can use Claude Code in a terminal, but even non-technical users can try it out using Claude's desktop app or web app.

For a full guide on getting started, check out our Claude Code guide.

Claude Cowork

With Cowork, accessed via Claude's desktop app, Claude can now do actual work on your computer. Just describe the outcome you want—like pulling data from a bunch of PDFs into a spreadsheet—and instruct Claude to handle it for you. Much like Claude Code, Cowork makes a plan, breaks it into steps, and does the work while you watch. It runs in a sandboxed environment, which means it can only take action within the folders you specify.

Cowork is perfect for admin-heavy tasks like organizing files, reconciling transactions, and streamlining onboarding. If you're used to procrastinating by organizing your desktop files and folders, you'll have to find a new pastime: Claude Cowork can do it in 30 seconds.

Claude Cowork also lets you schedule recurring tasks, like pulling sales data into a report each morning or reconciling expenses at the end of each week. And with Dispatch, another Claude feature, you can control Cowork sessions remotely from your phone (as long as your computer is turned on) so you can kick off a task from anywhere.

Claude Design

Claude Design lets you create prototypes, wireframes, pitch decks, landing pages, and marketing collateral, and it tends to create more polished results than you'd get from other Claude tools. You can refine your design within the chat or with inline comments, direct edits, or custom sliders. 

Claude Design can also automatically apply your team's design system across everything you build, so the output is consistent whether you're building a slide deck or an app prototype. Once you're happy with your design, you can share it via a URL, save it as a PDF or PPTX file, or export it to Canva. And if you want to take your design and build something with it, Claude Design can create a bundle of design files and instructions so you can continue building in Claude Code.

Claude Design is currently in research preview. You can access it with any paid Claude subscription.

How to try Claude for yourself

For access, sign up at Claude.ai. From there, you can start a conversation or use one of Claude's default prompts to get started. As a free user, you'll get limited access to Claude's Sonnet and Haiku models. Upgrading to one of Claude's paid plans (starting at $20/month) gives you access to additional models and priority access even during times of high traffic.

You can also download Claude's desktop app for macOS or Windows, which streamlines access to Claude Cowork and Claude Code.

Automate Claude with Zapier

If you decide to use Claude as your AI chatbot of choice, you can connect it to Zapier so you can safely kick off automated workflows across 9,000+ apps directly from your Claude chat. Zapier handles OAuth-managed authentication for every connected app, so Claude gets the access it needs to act without getting access to your credentials. Learn more about how to automate Claude with Zapier MCP.

Zapier also offers access to the latest Claude models inside its own workflow engine—so whether you're kicking off an action from Claude chat or running a scheduled workflow in the background, you're working with state-of-the-art AI.

Related reading:

• What is Google Gemini?

• What is Perplexity AI?

• What are Claude computer use and OpenAI Operator?

This article was originally published in September 2023. The most recent update was in June 2026.