How Much Does Cline Pass Cost and What Models Does It Include?
Cline Pass costs $9.99 a month for open models like Kimi K3.1 Pro, GLM 5.3 and DeepSeek V4.1 Flash, with 5-hour, weekly and monthly caps. What it covers.
How much does Cline Pass cost and what does it include?
Cline Pass costs $9.99 per month. It is a subscription from the Cline team that bundles several open models behind one plan, with usage limits applied over 5-hour, weekly and monthly windows. The team negotiated volume discounts with the AI labs to make that possible. The lineup includes Kimi K3.1 Pro (the default), MiniMax M3.1 Flash, DeepSeek V4.1 Flash, GLM 5.3, GLM 5.3 Flash and Qwen3 Max.
TL;DR
- Cline Pass is a flat $9.99 monthly plan that gives you a set of open models instead of per-token billing.
- Kimi K3.1 Pro is the default model, with GLM 5.3, GLM 5.3 Flash, Qwen3 Max, MiniMax M3.1 Flash and DeepSeek V4.1 Flash alongside it.
- Usage is capped across 5-hour, weekly and monthly windows. The sources describe these as generous for the price but give no exact numbers, so check Cline's current terms.
- The plan suits token-heavy workflows such as multi-agent review, where a $20 Claude or Codex plan can run out quickly.
- You can use it in the Cline VS Code extension, the Cline CLI, or in other agents through an API key and the Cline API base URL.
- Mixing models inside one subscription is a practical advantage, because reviewer agents benefit from not sharing the same blind spots.
What models are in the Cline Pass lineup?
The lineup was recently refreshed. Here is what the pass covers, along with how one user who runs it daily assigns each model.
| Model | Notes |
|---|---|
| Kimi K3.1 Pro | Now the default model. Used as the long-conversation, design-writing agent. |
| GLM 5.3 | Used for code-reading work such as design critique and implementation checks. |
| GLM 5.3 Flash | Lighter GLM variant. |
| DeepSeek V4.1 Flash | Fast, suited to lighter review passes. |
| MiniMax M3.1 Flash | Fast, also suited to lighter passes. |
| Qwen3 Max | Rounds out the set. |
The role assignments in the right column are one creator's preferences, not official guidance. Treat them as a starting point and test against your own codebase.
What are the usage limits?
Cline Pass uses three overlapping windows: 5 hours, a week and a month. A short burst of heavy work can hit the 5-hour cap, and steady heavy use over several days can hit the weekly one. The monthly cap is the backstop.
The source material calls the limits "really generous for the price" but does not publish quotas. Do not assume a specific number of requests or tokens. If you plan to run a workflow that fans out into many parallel agents, read the current limits on Cline's site and watch your own consumption during the first week.
Why do token-heavy multi-agent workflows fit this plan?
Multi-agent review is expensive by design. Each fresh agent starts with no memory, so it reads the whole document and the referenced code from scratch. Run five or six of those at once and token use climbs fast.
A good example is the Elephant Goldfish Model (EGM), an open-source plugin from a Google team, licensed Apache 2.0. The repo states that it is not an officially supported Google product. It is based on an essay by Dave Renson titled "Elephants, Goldfish, and the New Golden Age of Software Engineering." The idea uses two kinds of agents plus you:
- The elephant remembers the project context, requirements and decisions, and writes the design doc.
- The goldfish is a fresh sub-agent with no memory of the conversation. It sees only the design doc and the files it points to, and it reviews them.
- You make the trade-offs and approve the design.
The brainstorm command sends out six goldfish reviewers in parallel, each with a different lens. One is a contrarian. Another looks at cost, quota and token economics. The review command then runs five stages: a zoom-out review, a comprehension review, a hostile design critic, a second zoom-out pass, and an implementation readiness check. The repo itself warns that this produces a lot of feedback and that it burns far more tokens than simply telling an agent to build the feature.
That is the case where the plan arithmetic matters. On a $20 Claude or Codex plan, a few review rounds can hit your limits quickly. On a flat $9.99 plan you can run brainstorm, draft and two or three review rounds on the same feature without watching a meter on every file read.
Is Cline Pass better than a $20 Claude or Codex plan?
It depends on the job. The two are different products, so "better" is the wrong frame.
Cline Pass makes sense when:
- Your workflow spawns many sub-agents that each re-read the same material.
- You are comfortable with open models rather than a single lab's flagship.
- You want to choose different models for different roles.
A $20 Claude or Codex plan makes more sense when:
- You want the lab's own model and tooling specifically.
- Your usage is light enough that you never come near the caps.
- Your work is small and clear, such as a simple bug fix, where a heavy review process does not pay off anyway. The EGM repo says as much.
Nothing in the sources benchmarks these open models against Claude or Codex, so this article makes no quality comparison. Judge output quality on your own tasks.
Why mix models in a review workflow?
A goldfish is useful because it does not share the elephant's assumptions. If the goldfish is also a different model, it does not share the same habits or blind spots either.
A split like this works inside one subscription:
- Kimi K3.1 Pro as the elephant, holding the conversation and writing the design doc.
- GLM 5.3 for the hostile design critic and the implementation readiness check, since both involve reading real code.
- DeepSeek V4.1 Flash or MiniMax M3.1 Flash for lighter passes such as the comprehension review.
Doing the same across separate vendor accounts would mean multiple bills and multiple keys. Here it is one plan.
How do you set up Cline Pass?
- VS Code: install the Cline extension and pick Cline Pass during onboarding.
- CLI: install the Cline CLI through npm as a global package and choose Cline Pass during onboarding. If you already configured it in VS Code, the CLI picks that up automatically.
- Other agents: create an API key in your Cline account and point the agent at the Cline API base URL.
The source also notes Cline Pass models are available in Pandaode, a tool built around an architect agent that dispatches work to separate workers. That structure maps neatly onto the elephant and goldfish pattern, with read-only workers acting as the reviewers.
Can you run EGM inside Cline?
Not through an official install. The source says Cline is not on Google's official support list yet, so setup is more manual than the Claude Code or Codex route. It can work because Cline now supports skill files and sub-agents, and that is what EGM relies on. The skills are plain markdown, and the reviewers are persona files in an agents folder.
Practical steps:
- Clone the whole repo, not just the skills folder. The skills read reviewer personas and document standards at runtime, so copying only the skills breaks them.
- Point Cline at the skill you want to run.
- Make sure each reviewer runs as a fresh sub-agent. If reviewers run in the same chat, you lose the point of the exercise. EGM itself refuses to run a review if it cannot spawn fresh sub-agents.
For Claude Code and Codex, EGM has documented plugin install commands. Those do not apply to Cline.
Frequently Asked Questions
How much is Cline Pass per month?
It is $9.99 per month. That price covers the full set of included open models, subject to the 5-hour, weekly and monthly usage limits.
Which model is the default in Cline Pass?
Kimi K3.1 Pro is now the default. You can switch to GLM 5.3, GLM 5.3 Flash, Qwen3 Max, MiniMax M3.1 Flash or DeepSeek V4.1 Flash depending on the task.
Can I use Cline Pass outside the Cline extension?
Yes. Create an API key in your Cline account and use it with the Cline API base URL in another agent. The Cline CLI and VS Code extension also work, and the models are available in Pandaode.
Will Cline Pass run out during multi-agent work?
It can, because the limits are capped over three windows. The source describes them as generous but does not publish exact figures. Heavy parallel review is the most likely way to hit a cap, so monitor early usage.
Is a multi-agent review worth it for small changes?
Usually not. The EGM repo says that for a small, clear bug fix you should use your normal workflow, since design and review take time and use many more tokens.
How we source, write, and correct our articles. Editorial standards
