Cline is an open-source AI coding agent for VS Code, the terminal and, in beta, a desktop app for Mac, Windows and Linux. The agent is free for individual developers. You pay only for the AI models it calls, either at cost through a Cline account or with your own API keys from providers such as Anthropic, OpenAI, Google, OpenRouter or DeepSeek, and it can also run local models through Ollama or LM Studio. ClinePass costs $9.99 a month and covers open-weight models such as GLM 5.3, Kimi K3 and DeepSeek V4 Pro without separate API keys. Enterprise, priced on request, adds the JetBrains extension, SSO, central billing and role-based access.
Reviewed by EdGrows·Updated Oct 3, 2026This analysis is based on documentation, public user reports, and vendor materials — not yet on our own hands-on testing. How we rate
Cline has no price tag of its own. The agent is open source and free for individual developers, and your bill is whatever the AI models it calls happen to cost. That can be very little or quite a lot, depending on the model you choose and how long you let the agent work.
It started as a VS Code extension that could read a project, plan a change and edit files with your approval. It now runs in four places: the editor, the terminal, a desktop app in beta and, through an SDK, inside other products. Its GitHub repository shows around 70,000 stars, and Cline says more than 11 million developers use it.
Cline's plans on October 3, 2026. ClinePass is sold on a separate page.
| Open source | ClinePass | Enterprise | |
|---|---|---|---|
| Price | Free | $9.99 a month | Custom |
| How AI is paid for | Your own API keys, or at cost through a Cline account | Open-weight models included | Central billing, with limits on which providers people can use |
| Where it runs | VS Code, terminal, desktop beta | Cline's IDE extension and CLI | Adds a JetBrains extension |
| For teams | Community support | Not aimed at teams | SSO, SLA, roles, a team dashboard and authentication logs |
Bringing your own key is the classic way to use Cline. It connects to the big labs directly and to routing services such as OpenRouter or AWS Bedrock, so you can follow whichever model is best this month. Buying credits through a Cline account does the same job with one bill, and Cline says it charges for inference at cost.
ClinePass is the newer option. For $9.99 a month after a promotional period, it covers a list of open-weight models from Z.ai, Moonshot AI, DeepSeek, MiniMax, MiMo and Qwen. GLM 5.3, Kimi K3 and DeepSeek V4 Pro are among them. Cline promises generous quotas without publishing exact limits, and says a processing fee may apply. For anyone happy with open models, it's a cheap and predictable way to run an agent all month.
Agents use far more tokens than chat. Every step involves reading files, running commands and sending the growing conversation back to the model, so a long task on a top model adds up quickly. Three settings make the biggest difference.
The first is the model. Cline lets you choose a different model for Plan mode, where it explores and proposes changes without touching files, and for Act mode, where it edits. Pairing a strong reasoning model for planning with a cheaper one for the edits is a sensible default. The second is task size: small, focused tasks keep the context short and the cost down. The third is a spending limit at your provider, worth setting before you let an agent run long tasks.
Cline also recommends turning on checkpoints before Act mode, so you can roll the workspace back if an edit goes wrong.
Cline for Desktop is a standalone app for Mac, Windows and Linux, currently in beta at version 0.0.43. You point it at a workspace and talk to the agent without opening an editor. You can queue follow-up messages while it works, or edit an earlier message to fork the session and rewind the workspace. One history holds sessions from the desktop app as well as the CLI and the extension. Cline warns that new builds ship often and rough edges are expected.
The CLI brings the same agent to the terminal, which suits scripted jobs and remote machines. The SDK lets teams embed the agent runtime in their own tools.
Local models are where Cline's model-agnostic design pays off. Its documentation covers Ollama, LM Studio and Atomic Chat. You start the local server, pick the provider in Cline and choose a model. Your code never leaves the computer, and the only cost is hardware. Cline suggests 16 to 32 GB of RAM for small quantized models and 64 GB or more for larger ones, plus a Use Compact Prompt setting that trims what the agent sends. Expect local models to be slower and less capable on big refactors than frontier models in the cloud.
The Enterprise plan is priced on request. It adds the JetBrains extension and single sign-on, with an SLA and dedicated support. Admins get central billing and role-based access, a team dashboard with authentication logs, and the ability to limit which inference providers developers may use. Advanced configuration management and fine-grained permissions are marked as coming soon.
The closest comparison, Claude Code, comes with the $20 Claude plan and works in the terminal and IDE. Cursor is a full AI editor at $20 a month, and GitHub Copilot Pro costs $10 with its own agent mode. On the free side, Aider lives entirely in the terminal, though its releases have slowed, and Continue.dev is the other open-source option for editors. The AI coding tools guide puts all of them on one page.
Cline makes the most sense for developers who care about choice. You can change models whenever a better or cheaper one appears, run everything locally, or pay $9.99 for open models, and the agent itself stays free.
The software is. Cline's extension, CLI and desktop app are open source and free for individual developers, with no seat fees. What costs money is the AI: every request goes to a model you choose, billed at that provider's rates through your own API key or at cost through a Cline account. ClinePass is an optional $9.99 subscription for open-weight models.
Yes. Cline's documentation covers Ollama, LM Studio and Atomic Chat: start the local server, pick the matching provider in Cline's settings and choose a model. It suggests 16 to 32 GB of RAM for small quantized models and 64 GB or more for bigger ones, and recommends turning on Use Compact Prompt.
A Cline subscription for open-weight models at $9.99 a month after any promotional period. It covers models from Z.ai, Moonshot AI, DeepSeek, MiniMax, MiMo and Qwen, including GLM 5.3, Kimi K3 and DeepSeek V4 Pro, without separate provider accounts or API keys. Cline notes that an additional processing fee may apply.
Claude Code is Anthropic's own agent and comes with the $20 Claude plan, so its usage sits inside a subscription. Cline is open source and works with Claude, GPT, Gemini or open models, billed per use. People who already pay for Claude often stick with Claude Code; those who want to switch models or control cost request by request tend to prefer Cline.
Only on Enterprise. The free version runs in VS Code, in the terminal and as a desktop app. Cline lists the JetBrains extension among its Enterprise features, along with SSO, central billing and role-based access control.
The fast issue tracker for software teams, now built for agents too: free up to 250 issues, Basic $10 and Business $16 per user a month billed yearly.
The AI code editor most developers compare against, now owned by SpaceX: Pro from $16 a month yearly, agents, Projects and its own Composer models.
A free terminal coding agent that commits every change to git. It still works with almost any model, but its releases have slowed to a trickle.