Cline pricing, free plan and limits
Open source AI coding agent that shows every edit before it runs.
- Best for
- Backend engineers who want to choose their own LLM instead of being locked into one vendor's assistant
- Not for
- Developers who mainly want passive inline autocomplete
Should you use Cline?
Cline suits developers who want to pick their own model instead of one vendor's built-in AI, and who want to see and approve every file edit and command before it runs. Its Plan/Act workflow and MCP support connect it to more than 30 different LLM providers, not just Claude or GPT.
A Plan/Act workflow paired with unrestricted Model Context Protocol tool access, so it can reach any external system a task needs without a closed integrations list.
Model-agnostic BYOK and deep MCP support create switching-cost lock-in through workflow and configuration, not model exclusivity; Cursor and Windsurf compete on editor polish, which Cline does not prioritize.
Also for
- Open source maintainers and teams who need to review every automated change before it lands
- Enterprise engineering teams needing SSO, centralized billing, and role-based access for a coding agent
Show 3 more
- Developers building their own agent products who want an embeddable runtime rather than a closed IDE
- Backend/platform engineers who want model choice
- Open source maintainers reviewing every change
Also not for
- Non-technical users without an LLM API key
Main trade-off
No inline autocomplete and a chat-first workflow make it a worse fit than Cursor or Copilot for fast, passive single-line edits.
Cline pricing
- Top plan
- $9.99/mo
- Billing model
- Freemium
Cline is free and open source; you pay your chosen model provider directly at cost. ClinePass is a $9.99/month add-on with quota-based access to a rotating set of open-weight models and no separate API key needed. Enterprise (SSO, SLA, dedicated support) is custom-priced via sales.
Before you commit
- LLM token spend scales with usage and is billed separately by the provider, not Cline
- ClinePass's promotional rate may differ from its stated standard $9.99/month rate after the promo period ends
What Cline does on its own
- Scope of autonomy
- Semi-autonomous
- Human approval boundary
- Human in the loop
- Maturity
- Generally available
Safety controls
- Audit loggingSupported
- Human in the loopSupported
- Runs sandboxedNot supported
- Output filteringNot supported
What it can operate
- Web browsing
- Code execution
- File system
- Shell commands
- API calls
Where Cline runs
- Deployment
- Self-hosted · API · SDK · Desktop
- Where it runs
- Desktop · CLI · API · IDE plugin
- Memory
- Short term
- Context window
- Model-dependent (set by the connected provider)
Models it runs on
- Model-agnostic (BYOK)
- Claude (Anthropic)
- GPT (OpenAI)
- Gemini (Google)
- DeepSeek
- 30+ providers via OpenRouter and direct API
Systems it connects to
| System | Type |
|---|---|
| GitHub | Native |
| VS Code | Native |
| JetBrains | Native |
| Model Context Protocol | API |
| Slack | Plugin |
| Linear | Plugin |
Benchmark scores
- Terminal-Bench 2.0 (89 tasks, Feb 2026)
- 57 / 100
What Cline does
Cline is an open source AI coding agent, run as a VS Code extension, JetBrains plugin, CLI, or SDK, with 69,000+ GitHub stars as of September 2026. It works with any connected LLM to modify a codebase, execute shell commands, and control a browser via Model Context Protocol tools, pausing for approval before each step unless auto-approve is enabled.
Plan then Act
A read-only Plan mode lets Cline explore the codebase and propose a strategy before Act mode changes a single file, so nothing is touched until you approve the plan.
Model Context Protocol tools
Cline connects to any MCP server for databases, APIs, or internal systems, with no fixed cap on how many servers one task can call.
Bring your own model
Cline works with more than 30 LLM providers, including Anthropic, OpenAI, Google Gemini, AWS Bedrock, and local Ollama models, so switching models never means switching tools.
Terminal and browser control
Cline runs shell commands and reacts to live output from tests, dev servers, and deploys, and can drive a real browser to verify its own changes.
Show all 6 features
Reviewable diffs and checkpoints
Every proposed file edit renders as a diff, and task checkpoints let you roll back to an earlier state if an autonomous run goes wrong.
Multi-surface runtime
The same agent runtime ships as a VS Code extension, a JetBrains plugin, a terminal CLI, a native desktop app, and a Node.js SDK for embedding in other products.
Strengths and limits
Strengths
- Free and open source, with no markup on model usage: you pay your chosen provider directly at its own token rate.
- Backed by $32 million in Seed and Series A funding from Emergence Capital and Pace Capital, giving the open source project more runway than most community tools have.
- Shows a diff or command preview for every file edit and terminal action before it runs, rather than applying changes silently in the background.
- Supports Model Context Protocol servers with no artificial limit on how many a single task can call, reaching databases and internal APIs some closed IDE agents cannot.
Limits
- Has no inline autocomplete: Cline is entirely task-driven through chat, so it skips the passive as-you-type suggestions that Cursor and GitHub Copilot provide.
- Model costs are billed separately by your chosen provider and can exceed a flat-fee competitor's subscription for developers running many long, large-context tasks daily.
- Enterprise pricing for SSO, an SLA, and dedicated support is not published and requires contacting sales for a custom quote.
- Extension stability during long sessions with heavy browser automation has been reported as inconsistent, occasionally requiring a restart.
Cline alternatives
| Agent | Where it runs | Price | Maturity | Checked |
|---|---|---|---|---|
| Cline This page | Your machine · API | Free plan | Available | 27 Sep |
 CursorCursor's $20-a-month Pro tier buys a more polished all-in-one IDE; Cline stays the free, open source, model-agnostic option instead. | -- | -- | -- | -- |
 WindsurfWindsurf offers a maintained VS Code fork with its own agent loop built in; Cline lets you plug in any model and inspect every action it takes. | -- | -- | -- | -- |
 Claude CodePick Claude Code for Anthropic's own terminal agent tuned to its models; pick Cline if you want one agent that runs Claude, GPT, Gemini, or a local model interchangeably. | -- | -- | -- | -- |
Also compared
GitHub CopilotPick Copilot for passive inline autocomplete while you type; pick Cline for autonomous multi-step tasks with a full plan you review first.
DevinPick Devin for a fully managed, cloud-hosted engineer at enterprise pricing; pick Cline for a free, self-hosted agent you run on your own machine.
Cline facts
- Works offline
- No
- Minimum skill level
- Intermediate
- Budget fit
- Free
- Open source
- Yes
- Ships
- Weekly
- Last major update
- Maker founded
- 2024
- Total funding
- $32M Seed + Series A (Emergence Capital, Pace Capital)
What changed
Questions about Cline
How much does Cline cost in 2026?
Cline itself is free and open source: you install the extension, plugin, or CLI at no cost and pay your chosen model provider directly for tokens, with no markup from Cline. The optional ClinePass subscription runs $9.99 a month and unlocks 14 open-weight models without needing your own key, at roughly 2 to 5 times the usage of standard API rate limits. Enterprise adds single sign-on and dedicated support, but that tier's price is not published and requires a custom quote.
What do you get on Cline's free tier?
The free, open source tier includes the full VS Code extension, JetBrains plugin, CLI, MCP marketplace access, and multi-root workspace support, with no feature paywall or task-count cap. The only recurring cost is whatever your chosen LLM provider charges per token, which solo developers commonly report at $20 to $50 a month for regular use.
Which tools compete with Cline in 2026?
Cursor and Windsurf are the closest IDE-embedded competitors, both shipping as VS Code forks with a more polished single-file editing experience at the cost of Cline's model choice and transparency. GitHub Copilot fits passive inline autocomplete better than autonomous multi-step tasks. Devin targets fully managed, cloud-hosted engineering work at enterprise pricing instead of Cline's bring-your-own-key setup.
Cline or Cursor: which should you pick?
Cline is free and open source with a bring-your-own-key model, so every file edit and terminal command needs your approval before it runs; Cursor is a closed-source VS Code fork whose $20-a-month Pro tier bundles faster single-file completions into one subscription. Pick Cursor for a maintained, polished daily editor, and Cline if avoiding vendor lock-in and inspecting every autonomous action matters more.
What does it take to start using Cline?
Install the free VS Code extension from the marketplace or run npm i -g cline for the terminal version, then sign in through the built-in Cline provider for a quick, no-API-key setup or add your own Anthropic, OpenAI, or Google key. Open Plan mode to describe the task first, since every action needs explicit approval before Act mode runs it.
About Cline
Cline started in June 2024 as a project by developer Saoud Rizwan for an Anthropic hackathon, built under the name Claude Dev before its October 2024 rename, and is released under the Apache 2.0 license. The project raised $32 million in combined Seed and Series A funding in 2025, led by Emergence Capital and Pace Capital with participation from 1984 Ventures, Essence VC, and Cox Exponential. Rather than a chat-only assistant, Cline plans a task, edits files, runs terminal commands, and can drive a browser to check its own work, pausing for your approval at each step unless you turn on auto-approve.
Read the full background
Cline is deliberately model-agnostic: instead of shipping with one fixed LLM, it connects to whichever provider you choose, from Claude Sonnet 5 and GPT-6 Astra to DeepSeek V4 or a local model, so its behavior depends on the model behind it rather than a model Cline trained itself. A Plan/Act workflow separates a read-only planning phase, where Cline explores the codebase and proposes a strategy, from an Act phase that actually touches files, and Model Context Protocol support lets it call external tools such as databases, ticket trackers, or internal APIs mid-task. Checkpoints record task history so a run can be rolled back if an autonomous change goes wrong.
The agent suits backend engineers and teams working across the coding agents category who want to choose their own model rather than being locked into one vendor's coding assistant, and organizations that need to review or approve every automated change to a shared codebase. Its Enterprise tier, used in some form by companies including Samsung, Salesforce, and Visa according to Cline's own site, adds single sign-on, centralized billing, and role-based access for larger engineering teams running the agent at scale.
The company kept up a new minor release roughly every one to two weeks through September 2026, including a September 14 desktop app launch with parallel sessions and cron-scheduled runs. On its own cline-bench initiative, the team raised Cline's Terminal-Bench score from 47% to 57% across 89 real-world tasks in February 2026 by diagnosing and fixing failures directly, an improvement it reported putting the agent ahead of several rival harnesses on that benchmark.
Sources and guides
Official sources
- Cline official site Vendor
- GitHub repository GitHub
- Community Discord
Autonomy, guardrails, integrations and pricing on this page are read from Cline's own documentation. The verdict is Hokai's reading of them, not the vendor's claim.
Hokai guides that mention Cline
Lists that include Cline
More AI Agents on Hokai
OpenAI CodexClones your repo, writes the code, opens the pull request. No keyboard required.
OpenHandsOpen-source AI agent that plans, codes, tests, and ships pull requestsPromptlessAI teammate that drafts doc updates as pull requests when you ship.
QualGentAutonomous QA agent that tests iOS and Android apps like a human.
OpenCodeOpen source coding agent that works with any model, in terminal, desktop or IDE.Codex SecurityOpenAI's security agent that scans repos and drafts fixes.