Last reviewed:
What is Claude Code? Definition and CIO question
Claude Code is Anthropic's agentic coding tool, launched in preview in February 2025 and generally available in May 2025. It runs in the terminal, code editors, the desktop app and the browser. It implements an autonomous coding agent that reads, writes, and executes code via the Claude API. To distinguish from the Claude model itself: Claude Code is the tool, Claude the model that drives it.
Launched in limited preview in February 2025, generally available in May 2025 with the release of Claude 4, Claude Code has become in 2026 one of the most adopted agentic tools by development teams. Architecture: MCP-compatible CLI agent, capable of reading and modifying project files, executing shell commands, running tests, creating Git commits. Tool slash commands distinct from model prompts. The official built-in commands include /init (project initialisation with CLAUDE.md), /clear (clear context), /compact (compress history), /help, /permissions, /mcp and /usage (session cost and plan usage limits; /cost is an alias). Bundled skills complete the list: /code-review (review of changed code, alias /review) and /loop (re-run a prompt at a regular interval). Claude Code is included in all paid Claude plans: Pro (USD 20/month, USD 17 on annual billing), Max (from USD 100/month, in 5x and 20x versions), Team (2 to 150 users, Standard seat at USD 20/month or Premium seat at USD 100/month on annual billing, with five times more usage) and Enterprise.
Concrete example
Illustrative case. A French tech mid-cap (180 employees, 10 developers) compares two options in 2026. Option A: GitHub Copilot Business at USD 19 per seat × 10 = USD 190/month, with USD 19 of AI Credits included per seat; beyond that, consumption has been billed on usage since 1 June 2026. Option B: Claude Code via the Team plan with Premium seats at USD 100 × 10 = USD 1,000/month on annual billing, a differential of USD 810/month. Standard seats (USD 20) also include Claude Code, with a lower usage volume. The difference is justified only if it is measured: track PRs delivered per developer over 3 months. Public benchmark: +26% completed tasks with Copilot in the trials by Cui et al. (2025). Agentic autonomy (multi-file reading, planning, execution) pays off only if the team maintains review discipline. On low-sensitivity code. For high-confidentiality code, arbitration tips in favour of self-hostable alternatives.
FAQ
What is Claude Code, and how does it differ from the Claude model?
Claude Code is Anthropic's agentic coding tool, launched in preview in February 2025 and made generally available in May 2025. It implements an autonomous coding agent that reads, writes and executes code via the Claude API. The distinction is simple: Claude Code is the tool, Claude the model that drives it.
What can Claude Code actually do?
It is an MCP-compatible agent, usable in the terminal, code editors, the desktop app and the browser, that reads and modifies a project's files, runs shell commands, runs tests and creates Git commits. It has tool-specific commands, distinct from model prompts, such as /init, /clear, /compact, /permissions and /usage, plus bundled skills such as /code-review.
How much does Claude Code cost?
It is included in all paid Claude plans: Pro (USD 20/month, USD 17 on annual billing), Max (from USD 100/month), Team (Standard seat at USD 20/month or Premium seat at USD 100/month on annual billing, with five times more usage) and Enterprise.
Is Claude Code suitable for confidential code?
With caution. Claude Code sends your repository context to the Anthropic API: for code subject to professional secrecy, critical intellectual property or defence compliance, the trade-off tips towards a self-hosted open-source model. The tool is best justified on low-sensitivity code.
What does it take to get a real gain from it?
Agentic maturity and discipline. Claude Code offers high autonomy (shell execution, multi-file edits, commits); without pre-validation and code review, the productivity gain turns into accelerated technical debt. In the mid-cap example, ROI becomes positive after six months if the team keeps that review discipline.
See also
Further reading
Sources
- Claude Code overview, official Anthropic documentation, code.claude.com (continuously updated, accessed 5 October 2026). https://code.claude.com/docs/en/overview
- Introducing Claude 4 (general availability announcement of Claude Code), Anthropic, 22 May 2025. https://www.anthropic.com/news/claude-4
- Claude pricing, Anthropic (Claude Code included in all paid plans), claude.com, accessed September 2026. https://claude.com/pricing
- GitHub Copilot is moving to usage-based billing, GitHub Blog, 2026 (Business $19, Enterprise $39 per user per month, AI Credits from 1 June 2026). https://github.blog/news-insights/company-news/github-copilot-is-moving-to-usage-based-billing/