Claude vs Claude Code: Which to Pick

Claude is Anthropic's assistant: you chat with it to write, research, analyze files and build things like documents and slides. Claude Code is an agent that reads your codebase, edits files, runs commands and works with your development tools. They use the same model family and, on paid plans, the same usage pool. Claude Code is included from Pro upward and not on the Free plan. Facts below were checked on 31 October 2026.

The short version

What each one is

Claude

Claude is the assistant and the family of models behind it. On the pricing page the Free plan lists chat on web, desktop and mobile, web search, file creation and code execution, memory across conversations, connected apps and tools, Artifacts, and Claude Design, Slides and Docs. Pro adds more usage, Claude Code, Projects, research, Claude in Chrome, and Claude for Microsoft 365, among other things.

The model line-up on the pricing page names Mythos (including Fable), Opus, Sonnet and Haiku. The Free plan includes Sonnet and Haiku. Opus arrives with Pro. Fable is paid with usage credits on Pro and included up to half your weekly limit on Max and Ultra.

Claude Code

Anthropic's docs: an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools. It runs in the terminal, in VS Code and JetBrains, in a desktop app and on the web at claude.ai/code. All surfaces use the same engine, so a repo's CLAUDE.md, settings and MCP servers work across them.

The product page lists recent additions: Code Review (generally available for Team and Enterprise), Projects in the desktop app to supervise several agents at once, auto mode by default on Pro, Max and Team plans, and self-hosted environments in public beta so sessions run on your own infrastructure.

Side-by-side facts

One pool of usage

This is the point many buyers miss. The pricing FAQ says that every plan has usage limits, that most accounts have a limit that resets on a rolling five-hour session window, and that paid plans add weekly limits on top. It then says your activity across Claude on web, desktop and mobile, and Claude Code, all draws from the same pool. There is no fixed message count, because how much you can do depends on the length and complexity of conversations, the model you choose, and the features you use.

In practice this means a long coding session can leave less for chat that day, and a day of heavy document work leaves less for code. If you find yourself hitting limits, you have three options:

1. Move up a plan. Max gives 5x Pro usage every five hours and every week. Ultra gives 10x the weekly usage with no five-hour limit for verified accounts.
2. Choose a lighter model for routine work.
3. Switch heavy coding sessions to pay-as-you-go API credits through a Console account. Max and Ultra come with monthly API credits ($100 and $200) for the Claude Platform.

Claude Code has a /usage screen that shows plan usage bars and a breakdown of what counts against your limit: skills, subagents, plugins and individual MCP servers, each as a percentage, with suggestions to trim.

Plans and prices in detail

Individual

Teams and enterprises

The Team plan is for teams of 2 to 150. A Standard seat is $20 per seat a month billed annually ($25 monthly), and a Premium seat with 5x the usage of a standard seat is $100 ($125 monthly). The product page says Team includes Claude Code for every seat, with self-serve seat management and up to $500 a month in pooled API credits. Enterprise is $20 per seat a month plus usage billed at API rates, billed annually, and adds SCIM, audit logs, custom data retention and role-based access.

Using the API directly

Claude Code does not require a subscription. It can sign in with an Anthropic Console account or an API key, and then you pay per token. Anthropic's cost page says that across enterprise deployments the average is around $13 per developer per active day, or $150 to $250 per developer per month, with 90 percent of users below $30 per active day. Those are averages from Anthropic, not a forecast for your project.

What Claude Code adds to Claude

The honest list, from the docs:

• A loop with tools. It reads files, edits them, runs commands and checks results, rather than describing code in a chat window.
• Project memory. CLAUDE.md is read at the start of every session. If your repo already has AGENTS.md, Claude Code can read that instead. It also builds auto memory as it works.
• Permissions. It asks before editing files or running shell commands, except a built-in set of read-only commands. Plan mode reads without editing. Bash rules match the whole command and understand operators.
• Skills. A SKILL.md file adds a repeatable workflow. Custom commands were merged into skills.
• Subagents. Specialized assistants with their own context window, system prompt, tools and permissions.
• Hooks. Shell commands, HTTP endpoints, MCP calls, prompts or subagents that run at points in the agent's lifecycle, the same in every surface.
• MCP. The Model Context Protocol connects it to design docs, tickets, chat and your own tools.
• Scripting and CI. claude -p for pipes, plus GitHub Actions and GitLab CI/CD.
• Scheduling. Cloud scheduled tasks keep running when your computer is off.
• Handoff. Remote Control, claude --teleport and /desktop move work between surfaces.

Most of those have no counterpart in a chat window.

What Claude has that Claude Code does not

• Artifacts, Design, Slides and Docs for non-code outputs.
• Voice mode, mobile apps and incognito chats.
• Research and web search as first-class features.
• Claude for Microsoft 365 and Google Workspace connections.
• A Free plan.

Claude Code can do some of the same things through MCP and skills, but the chat product is where those features live.

Install and first run

Claude Code on macOS, Linux or WSL:

curl -fsSL https://claude.ai/install.sh | bash
claude --version
cd your-project
claude

On Windows, the docs give a PowerShell command (irm https://claude.ai/install.ps1 | iex) and a CMD version. Homebrew has two casks: claude-code tracks the stable channel, typically about a week behind, and claude-code@latest gets releases right away. WinGet is winget install Anthropic.ClaudeCode. You can also use apt, dnf or apk. Native installs update in the background; Homebrew and WinGet installs do not.

You do not need to install anything for Claude itself. Open claude.ai, or download the apps. The desktop app includes Claude Code under its Code tab, so you don't need the CLI separately, and a paid subscription is required.

Models

Both draw on the same models. Claude Code's model docs describe aliases: opus, sonnet, haiku, fable and best, where best uses the model fable resolves to where available, otherwise the same as opus. An opusplan mode plans with Opus and executes with Sonnet. The aliases resolve to the newest version on the Anthropic API and to an earlier version on some other providers.

Claude Code in the terminal, VS Code and JetBrains also supports third-party providers, such as Amazon Bedrock and Google Vertex AI.

Data handling

Anthropic's data-usage page treats consumer plans and commercial plans differently. For Free, Pro, Max and Ultra accounts, there is a setting that allows your data to be used to improve future models, and it applies to Claude Code use from those accounts too. Users who allow it face a five-year retention period. For Team and Enterprise plans, the API and third-party platforms, Anthropic says it does not train generative models on code or prompts sent to Claude Code under commercial terms, unless the customer opts in, for example through the Developer Partner Program.

The page also notes that feedback sent from inside Claude Code is only uploaded if you explicitly choose Yes, and that responses to its survey do not change your training preference.

If you use both products on a personal plan, the one training setting covers both. Check it before opening a private repository.

Which to pick

Use Claude (chat) when:

• The work is writing, research, analysis or documents.
• You want to try Anthropic at no cost.
• You want mobile, voice or Microsoft 365 connections.

Use Claude Code when:

• You change code in a repo often.
• You need an agent in CI, scheduled, or run on your own infrastructure.
• You want a team to share hooks, skills and subagents.

Pay for a plan and use both when:

• You code in the morning and write in the afternoon. Pro at $20 covers both, subject to one shared usage pool.

Common questions about the split

Is Claude Code a different model? No, it is an agent that uses Claude models.

Can I run it without the chat product? Yes, with a Console account or API key.

Does Max help only with code? No. Max raises the usage pool for everything, because the pool is shared.

Do Team seats include Claude Code? The product page says Team includes Claude Code for every seat.

A short checklist

• Look at your usage page after a week. Which product eats the pool?
• Decide whether you need Max for the extra headroom or an API account for bursts.
• Set the training preference deliberately.
• Add a CLAUDE.md before the first long Claude Code session.
• Put Claude Code in CI only after you review its permissions.

Our take

Going by the documented facts, this is less a choice than a ladder. Claude is the front door, and Claude Code is a specialist tool behind the paywall, drawing on the same pool of usage. If you are new, start with Claude on the Free plan to find out whether the model suits your work. When you want an agent to touch your files, step up to Pro and add Claude Code.

The one decision that deserves thought is how you pay for heavy coding. A single pool is simple until a long agent session crowds out your chats. Max or API credits fix that, at a price you can read off the pricing page. We did not measure how fast a typical coding session uses the pool, so this page does not give a number.

Related

• Claude Code
• ChatGPT vs Claude
• Cursor vs Claude Code
• Claude Code vs Codex
• MCP servers
• How we compare