Comparison
Cline vs Claude Code: the open-source agent against Anthropic's
One is a harness for any model you bring; the other is Anthropic's own agent with Anthropic's plans. Where each wins, and whether you can have both.
Cline is a free, open-source coding agent (Apache 2.0, about 70,000 GitHub stars) that runs in VS Code, JetBrains, a terminal CLI and a desktop app, with any model you choose: you bring an API key, buy usage from Cline at cost, or pay $9.99 a month for ClinePass on open-weight models. Claude Code is Anthropic's own agent, included in Claude Pro from $20 a month and running Claude models only. Claude Code is the more proven of the two, first on Terminal-Bench 4.0 with Opus 5.5 and far more widely installed; Cline isn't on the independent leaderboards. Cline can also drive your installed Claude Code with a Claude subscription, or use a ChatGPT subscription for OpenAI's models.
Key takeaways
- Price: Cline is free; you pay for the model. Claude Code comes with Claude plans: Pro $20 ($17 annually), Max $100 or $200.
- Models: Cline works with Anthropic, OpenAI, Google, open-weight and local models; Claude Code uses Claude only.
- Approvals: Cline asks before each edit and command in the editor, but its CLI auto-approves by default; Claude Code starts in auto mode with a safety classifier.
- Adoption: about 5.6 million VS Code installs for Cline against 27 million for Claude Code's extension, on October 7.
- Benchmarks: Claude Code with Opus 5.5 leads Terminal-Bench 4.0 at 64.8%; Cline has no independent result.
| Cline | Claude Code | |
|---|---|---|
| Made by | Cline Bot Inc., open source (Apache 2.0) | Anthropic |
| Price | Free; pay for tokens, or ClinePass $9.99 | Claude Pro $20 and up, or API |
| Models | Any: Claude, GPT, Gemini, open-weight, local | Claude only |
| Where it runs | VS Code, JetBrains, CLI, desktop, cloud | Terminal, VS Code, JetBrains, desktop, web, phone |
| Approvals | Per action in the editor; CLI auto-approves | Auto mode with a classifier, or manual |
| VS Code installs (Oct 7) | 5.6 million | 27.2 million |
| Terminal-Bench 4.0 | Not listed | 64.8% with Opus 5.5 (first) |
| Uses a Claude subscription | Through your installed Claude Code | Yes |
What is Cline?
An open-source agent from Cline Bot Inc., licensed Apache 2.0 (GitHub). In 2026 it rebuilt itself on its own SDK and now comes as:
- Editor extensions for VS Code (also Cursor, Antigravity and other forks through Open VSX) and JetBrains.
- A CLI (
npm i -g cline), version 3 since May 2026, scriptable with JSON output and schedules. - A desktop app in beta since September 2026, and Cline Cloud, which runs tasks in a cloud sandbox and opens pull requests.
It has Plan and Act modes (Plan "cannot modify any files or execute commands"), checkpoints you can restore after every step, MCP servers, rules in .clinerules/ (it also reads AGENTS.md), skills and hooks (Cline).
How much does Cline cost compared with Claude Code?
- Cline: the extension, CLI and desktop app are free. You pay for tokens with your own key, or through Cline's provider at cost. ClinePass is $9.99 a month for "2-5x the usage" on open-weight models such as Kimi K3 and GLM-5.3, with limits that reset every five hours (Cline). Teams features are Enterprise-only, at a custom price.
- Claude Code: Pro $20 a month ($17 billed annually), Max $100 or $200, Team seats from $20, or per token with an API key (Claude Code pricing).
Running Claude Sonnet 5.5 in Cline with an API key costs Anthropic's list price, $2 per million input tokens and $10 output; a Claude plan is usually cheaper for daily use.
Can you use a Claude Pro or Max subscription in Cline?
Yes, through Cline's "Claude Code" provider: it runs the Claude Code program you've installed and signed in to, rather than using your login itself (Cline). Anthropic's terms bar apps that sign in with your Claude account on their own, but say they don't prevent "an end user from signing in to the unmodified Claude Code binary with their own Claude subscription" (Anthropic). That's how Cline's route works, though Anthropic hasn't commented on Cline specifically. Cline also has a "Sign in with OpenAI" option for ChatGPT plans.
Is Cline or Claude Code better?
On the evidence available, Claude Code:
- Terminal-Bench 4.0 (October 6): Claude Code with Opus 5.5 is first at 64.8%; Cline isn't listed (tbench.ai).
- Artificial Analysis's coding agents index: Claude Code with Sonnet 5.5 scores 68; Cline isn't included (Artificial Analysis).
- Cline's own figures: it reports 82% on the older Terminal-Bench 2.0 with Kimi K3, run by Cline itself.
Cline's case is flexibility: any model, including cheap open-weight and local ones, inside the editor you already use, with every step approvable and restorable. Claude Code's is depth: Anthropic tunes it for Claude, and it adds cloud sessions, a phone app, skills, hooks and agent teams.
Cline vs Claude Code in VS Code
Both install from the VS Code Marketplace. Cline shows each step in a side panel and, by default, asks before each file edit and command; you can auto-approve by category, and command auto-approval is off by default since version 4. Claude Code's extension shows side-by-side diffs and plans you can comment on, and starts in auto mode (How to use Claude Code). On the command line, note that Cline's CLI approves everything unless you pass --auto-approve false.
Cline vs Codex
Codex is OpenAI's own agent, open source and included with ChatGPT. Cline can run on the same ChatGPT subscription through its OpenAI provider, so the choice is about the harness, not the bill (How to use Codex). More options in Claude Code alternatives.
Cline and Claude Code for teams
Cline Enterprise adds SSO, roles, central billing, provider and MCP allowlists, and shared rules; Claude Code shares settings, skills and hooks through the repository and has Team and Enterprise plans (Claude Code for teams). Neither lets several people steer and approve one agent together. Poly (usepoly.co) does: a team works with one Claude Code or Codex agent in a shared browser room, everyone sees each step live, and any member can approve a change before it happens. Free to start. What is Poly?
Common questions
Is Cline better than Claude Code?
On independent benchmarks Claude Code leads: it is first on Terminal-Bench 4.0 with Opus 5.5, and Cline isn't listed. Cline is better if you want any model, including cheap open-weight or local ones, in an open-source tool.
Is Cline free?
The Cline extension, CLI and desktop app are free and open source. You pay for the model you use, with your own API key, through Cline's provider at cost, or with ClinePass at $9.99 a month for open-weight models.
Can I use my Claude subscription with Cline?
Yes. Cline's Claude Code provider runs your installed, signed-in Claude Code, so it uses your Pro or Max plan. Anthropic's terms allow signing in to the unmodified Claude Code program with your own subscription.
What is the difference between Cline and Claude Code?
Cline is an open-source agent that works with any model and asks before each action in your editor. Claude Code is Anthropic's agent for Claude models, with plans from $20, cloud sessions, a phone app and team plans.
Cline or Codex?
Codex is OpenAI's own agent, included with ChatGPT. Cline can use the same ChatGPT subscription through its OpenAI provider, so the choice comes down to which tool you prefer to work in.