Claude Code
Claude Code is Anthropic's terminal-based agentic coding CLI, installed via `npm install -g @anthropic-ai/claude-code` or a native installer; version 2.1.210 shipped July 14, 2026 (prior build 2.1.207, July 11). It's proprietary software under Anthropic's Commercial Terms of Service, not an OSI-approved license, though the repo (github.com/anthropics/claude-code) is publicly readable on GitHub. It's bundled into paid Claude plans β Pro $20/mo, Max 5x $100/mo, Max 20x $200/mo β sharing one usage pool across chat and CLI, gated by 5-hour rolling limits plus separate weekly all-model and Sonnet-only caps; also billable standalone via API around $1-$25 per million tokens depending on model. Anthropic reported Claude Code's run-rate revenue topped $2.5B by February 2026, and a Pragmatic Engineer developer survey put its coding-tool share at 46% versus Cursor's 19% and GitHub Copilot's 9%. Current version/status: 2.1.210 (July 14, 2026). License: Proprietary β Anthropic Commercial Terms of Service (repo publicly readable, not OSI-licensed). Pricing: Bundled into Claude Pro $20/mo, Max 5x $100/mo, Max 20x $200/mo, and Team Premium ~$100/seat/mo (shared usage pool with 5-hour rolling + weekly caps); also usable standalone via API pay-as-you-go at roughly $1-$25 per million tokens depending on model. Maintained by Anthropic PBC.
Pick Claude Code when you live in a terminal and want an agent that autonomously plans multi-file refactors and CI-scriptable workflows at high effort against Sonnet 5/Opus 4.8 with a 1M-token window β reviewers consistently note it burns fewer tokens per task than Cursor's chat-heavy loop. Skip it if you want a polished IDE diff/review UI (Cursor still wins there) or refuse a subscription lock-in β Aider is the honest pick: free, Apache-2.0, model-agnostic, bring-your-own-key. Even Anthropic's own docs concede the catch: Max plans impose both a 5-hour rolling cap and separate weekly all-model/Sonnet-only limits that reset on a fixed clock, not your usage pattern, so heavy users routinely get throttled mid-task. Known weakness: Even top-tier Max plans impose a 5-hour rolling session limit plus separate weekly all-model and Sonnet-only usage caps that reset on a fixed account-assigned clock rather than actual usage pattern, so heavy users regularly hit throttling mid-task and must upgrade tiers or wait out the reset.
See how it ranks β