Both cost $20/mo. The difference is what they do.
Anthropic's terminal coding agent
AI-first IDE, multi-file agent edits
They overlap more than they used to: both run subagents, MCP, hooks and skills, both have a CLI and cloud agents, and the Claude Code extension installs inside Cursor. The split is the editor and the meter. Cursor is an IDE that sells a monthly usage budget you can spend on its own Grok and Composer models or on Claude, GPT and Gemini. Claude Code is Anthropic's agent, included with Claude Pro and up, metered in five-hour windows on Anthropic models only.
| Attribute | Claude Code | Cursor |
|---|---|---|
| What you buy | An agent, included in a Claude subscription or billed per token on the API | An IDE with a monthly model usage budget, plus a CLI and cloud agents |
| Entry price | No free tier; Claude Pro, $20/mo or $17/mo billed annually | Free Hobby with limited Agent requests, then Pro at $20/mo |
| Higher individual plans | Max 5x $100/mo, Max 20x $200/mo | Pro+ $60/mo, Ultra $200/mo |
| Team seats | Team Standard $25/mo, Premium $125/mo ($20 and $100 billed annually) | Teams Standard $40/user/mo, Premium $120/user/mo with 5x the Standard agent usage |
| Default model | Sonnet 5 on Pro and Team Standard; Opus 5 on Max, Team Premium, Enterprise and the API | Auto picks a model per request; Cursor calls Grok 4.5 its flagship and Composer 2.5 its fast model |
| Model choice | Anthropic only: Sonnet 5, Opus 5, Fable 5.1 and Haiku | Grok 4.6, Grok 4.5 and Composer 2.5 in the Cursor pool; Claude Opus 5, Sonnet 5, Fable 5.1, GPT-5.6 Sol, Terra and Luna, Gemini 3.1 Pro and others at API rates |
| Subagents | Yes: custom subagents in their own context, run in the background by default; agent teams experimental | Yes: built-in Explore, Bash and Browser subagents, custom ones in .cursor/agents, foreground or background, plus cloud subagents with /in-cloud |
| Rules and extensions | CLAUDE.md, .claude/rules, hooks, skills, plugins, MCP; /init imports .cursor/rules | .cursor/rules, AGENTS.md, user and team rules, hooks, skills, plugins, MCP |
| Where it runs | Terminal, VS Code and Cursor extension, JetBrains, desktop app, web, the Claude mobile app, Slack, GitHub Actions and GitLab CI/CD | Its own editor and Agents Window, a CLI, and cloud agents from the web, iOS, Slack, GitHub, Bitbucket, Linear and an API |
| Usage meter | Shared with Claude chat: a session limit that resets every 5 hours plus a weekly limit; usage credits or an API key beyond that | Two monthly pools, Cursor Models and Other Models, that reset with the billing cycle; opt-in on-demand usage at API rates once they run out |
They overlap more than they used to: both run subagents, MCP, hooks and skills, both have a CLI and cloud agents, and the Claude Code extension installs inside Cursor. The split is the editor and the meter. Cursor is an IDE that sells a monthly usage budget you can spend on its own Grok and Composer models or on Claude, GPT and Gemini. Claude Code is Anthropic's agent, included with Claude Pro and up, metered in five-hour windows on Anthropic models only.
Cursor has the lower floor. The Hobby plan is free with limited Agent requests, and the first paid plan is Pro at $20/mo. Claude Code is not on Claude's Free plan and starts with Claude Pro at $20/mo, or $17/mo billed annually. Above that, Cursor sells Pro+ at $60/mo and Ultra at $200/mo, and Claude sells Max 5x at $100/mo and Max 20x at $200/mo. For teams, Claude Team Standard costs $25 per seat billed monthly or $20 annually against $40/user/mo for Cursor Teams Standard, while the premium seats are close: $125/mo or $100 annually for Claude, $120/user/mo for Cursor. Cursor's own guidance is that daily agent users spend $60 to $100 a month in total usage and power users $200 or more.
Yes. Anthropic's Claude Code extension for VS Code has an install link for Cursor and also installs in other VS Code forks, and if an editor cannot install it, the CLI runs in the integrated terminal. Claude Code then bills to your Claude plan or API key, not to Cursor, while your Cursor plan still covers Tab, Cursor's own agent and Bugbot. Claude Code's /init command reads existing .cursor/rules and .cursorrules files when it generates CLAUDE.md, so a repo set up for Cursor carries over. The reverse works too: pick Claude Opus 5, Sonnet 5 or Fable 5.1 in Cursor's model picker and they draw on Cursor's Other Models pool at Anthropic's API rates, though Fable 5.1 needs data retention approval when Privacy Mode is on.
They meter differently. Claude Code shares limits with Claude chat: Pro and Max have a session limit that resets every five hours plus a weekly limit, Team and Enterprise seats draw from a per-seat allowance on the same rolling five-hour and weekly windows, and usage credits let you keep going past the cap. An API key bills per token instead. Cursor meters a monthly budget: Pro, Pro+ and Ultra include two usage pools that reset with your billing cycle, a Cursor Models pool for Grok 4.6, Grok 4.5 and Composer 2.5, and an Other Models pool for third-party models charged at their API price. When a pool runs out you either enable on-demand usage at the same API rates with no markup, under a spend limit you set, or upgrade. Teams and Enterprise plans add a Cursor Token Rate of $0.25 per million tokens on third-party models.
Claude Code runs Anthropic models only: Sonnet 5 by default on Pro and Team Standard, Opus 5 by default on Max, Team Premium, Enterprise and the API, with Fable 5.1 and Haiku selectable. Cursor's Auto setting picks a model per request, or you choose one: Cursor's own Grok 4.6, Grok 4.5 and Composer 2.5 draw on the Cursor Models pool, and Claude Opus 5, Sonnet 5 and Fable 5.1, GPT-5.6 Sol, Terra and Luna, Gemini 3.1 Pro, Gemini 3.8 Flash, Muse Spark 1.3 and others draw on the Other Models pool at each provider's API price. On Teams and Enterprise, Cursor Router routes Auto requests across GPT-5.5, Claude Opus 5, Grok 4.5 and Claude Fable 5.1. Neither tool runs local models.
Anthropic's terminal coding agent. It sits in the coding agents category, and ships in 4 tested stacks here.
AI-first IDE, multi-file agent edits. It sits in the ai ides category, and ships in 6 tested stacks here.
They are alternatives at the same step, so most stacks pick one or the other. No aistack.sh stack uses both today, though nothing prevents a custom workflow.
Beyond each other, Claude Code's closest alternatives are Codex. Cursor's closest alternatives are Codex.
