Cline and Kilo Code vs Claude Code and Codex in 2026: open-source agents, BYOK, and which subscription still works
Cline and Kilo Code are the two big open-source coding agents left after Roo Code shut down in May. Both are free and bring-your-own-key, but a Claude Pro or Max login no longer works in Kilo, ChatGPT sign-in works in both, and the gateways have their own fine print. We read their pricing pages, docs and Anthropic's and OpenAI's rules to show who pays for each run, and where the brief and design states should live.

Open-source agents got cheaper, and the subscription rules got stricter
A year ago the open-source coding agent scene had three big names: Cline, its fork Roo Code, and Kilo Code, a fork of Roo. Roo Code shut down its extension, cloud and router on 15 May 2026. Kilo Code was bought by Anaconda in July and rebuilt its version 7 on the OpenCode server. Cline passed 70,000 GitHub stars and shipped a CLI, a desktop app and cloud agents. Both are free to install, and both let you bring your own model key.
Over the same months the rules about subscriptions changed. In January 2026 Anthropic restricted Claude Pro and Max credentials to its own clients, and its legal page now says third parties may not offer Claude.ai login or route requests through Free, Pro, Max or Ultra plans. OpenAI went the other way: since DevDay on 29 September, Sign in with ChatGPT lets partner apps spend your plan's included Codex allowance.
So the question for a small team is no longer which agent is best. It is who pays for each run, and how many meters you are running at once. Most product teams already pay for six tools with six AI allowances, and the brief, the states and the decisions sit across all of them. This guide compares Cline and Kilo Code with Claude Code and Codex from vendor pages checked on 22 November 2026.
How Cline charges in November 2026
Cline's pricing page now has two tiers, down from three. There is no per-seat price for individuals or small teams.
- Open Source, free. The VS Code extension, the CLI and the MCP Marketplace. Inference is bought at cost through Cline, or you bring your own key from Anthropic, OpenAI, OpenRouter, a cloud provider or a local model.
- ClinePass, $9.99 a month. An optional plan for curated open models such as DeepSeek, GLM, Kimi, Qwen and MiniMax, with two to five times the API rate limits.
- Enterprise, custom price. Adds the JetBrains plugin, SSO, RBAC, audit logs, central billing, limits on which providers people may use, and an SLA. The Teams tier that was free up to ten users and then $20 per user is no longer on the page.
- Subscriptions. Cline has had OpenAI Codex sign-in since 22 January 2026, so a ChatGPT plan pays for runs. Its docs also list a Claude Code provider that drives the claude CLI installed on your machine; check Anthropic's current terms before you rely on it for a team.
- Rules. Cline reads .clinerules first and also reads AGENTS.md files, including nested ones and a global file in your home folder.
Cline's appeal is control. Plan and Act modes, checkpoints you can roll back, a Kanban for parallel agents in git worktrees, and every model you can name. The cost is attention: with BYOK, nobody caps your spend but you.
How Kilo Code charges in November 2026
Kilo Code sells a team seat and a gateway, and keeps the extension free.
- Individual, free. VS Code, JetBrains, the CLI and Kilo Desktop. Usage is billed separately through your own key or Kilo credits.
- Teams, $15 per user per month. A 14-day trial, usage analytics, shared custom agents, shared BYOK keys and central billing. Enterprise adds SSO, SCIM, audit logs and EU inference.
- Kilo Gateway. Over 500 models at what Kilo calls exact provider rates with no markup, but a 5% processing fee applies when you buy credits, on every tier.
- Kilo Pass, $19, $49 or $199 a month. A prepaid credit plan that adds bonus credits on top of what you pay.
- Cloud runs. Cloud Agents cost $0.60 or $1.20 an hour of compute and Code Reviews $0.33 an hour, billed per second, on top of tokens.
- Subscriptions. ChatGPT Plus and Pro sign-in works in the extension, the CLI, cloud agents and code reviews. Claude Pro and Max do not: Kilo's own docs say Claude subscriptions cannot be used in third-party harnesses, and its Claude linked billing works only with a Claude API organization billed per token.
Kilo's version 7 simplified its modes to code, ask, plan and debug. Architect became plan, and Orchestrator is deprecated because agents now start subagents on their own.
Cline, Kilo Code, Claude Code and Codex side by side
All four read a repository, plan, edit files, run commands and open a pull request. They differ in who sells the model access and which plan you can sign in with.
| Agent | Plans | How a run is paid | Rules file and context |
|---|---|---|---|
| Cline | Free; ClinePass $9.99; Enterprise custom | BYOK, Cline credits at cost, or ChatGPT sign-in | .clinerules, AGENTS.md, MCP Marketplace |
| Kilo Code | Free; Teams $15 per user; Kilo Pass $19 to $199 | BYOK, Kilo credits plus 5% fee, or ChatGPT sign-in; no Claude plan login | AGENTS.md, custom agents, MCP |
| Claude Code | Claude Pro $20, Max $100, Ultra $200 | Your Claude plan's usage limits; API key billed per token | CLAUDE.md, any MCP server you add |
| Codex | ChatGPT Plus, Pro, Business plans | Your ChatGPT plan's Codex allowance; API key billed separately | AGENTS.md, MCP servers in config.toml |
| Sydla workspace | $19 per user | No AI credits: drives your own Claude Code or Codex, zero token markup | Brief, criteria, UX Studio states and evidence in one MCP connection |
Read the table by the plan you already pay. If an engineer has a ChatGPT plan, Codex, Cline and Kilo Code can all spend it. If they have a Claude plan, Claude Code is the supported way to use it, and Cline or Kilo would need an API key on a second, per-token bill. Our bring your own coding agent guide explains the three ways to pay for a run, and the Claude Code vs Codex guide compares the two plans.
The six-tool stack around any agent
Whichever agent you pick, it still needs a ticket, a brief, designed states, decisions and a launch visual. In a typical 2026 stack those sit in six tools, each with its own seat and AI meter, and each one an MCP hop away from the run.
| Tool | Seat price | Bundled AI and its limit | Handoff to a coding agent |
|---|---|---|---|
| Jira Standard | $9.70 | Rovo: 25 credits per user per month, pooled; PAYG $0.01 per credit from 3 Dec 2026 | Atlassian MCP or a pasted ticket |
| Confluence Standard | $7.15 | Same Rovo pool, 25 credits per user | Atlassian MCP; enriched calls 1 to 10 credits |
| Notion Plus | $12.00 | Custom Agents on Business and Enterprise; extra credits $10 per 1,000 | Notion MCP server |
| Figma Professional Full | $20.00 | 3,000 AI credits per Full seat, no rollover | Figma MCP: 200 calls a day on Full or Dev |
| Miro Starter | $10.00 | 25 AI credits per license, pooled per team | Miro MCP: daily call limits by plan |
| Canva Business | $25.00 | Personal AI allowance per member | Outside the build loop |
| Sydla workspace | $19.00 | No AI credits: bring your own Claude Code or Codex | One MCP connection or a Sydla Desktop thread |
The six tools come to $83.85 per user per month before any overage, against $19 for Sydla, a saving of $64.85. Add Kilo Teams for the agent seat and the stack is near $100 per person before a single token. The stack cost guide works through 5, 10 and 25 people, and the comparison overview covers each tool.
Technical walkthrough: your own agent on a shared brief and UX Studio states
Sydla does not add another agent or another gateway. The context lives in one workspace, and the agent you already pay for reads it on your machine.
- Brief. Each ticket sits in an initiative with a versioned brief, numbered acceptance criteria and a done-means line. Every save is signed by its author, a person or an agent.
- UX Studio. Screens and their states, empty, loading, error and success, are designed next to the ticket and published as a handoff with tokens and a URL per state.
- Run. In Sydla Desktop you open the ticket, start a thread and choose claude or codex. Desktop drives the CLI you are already signed in to, so the model, the limits and the bill stay on your Claude or ChatGPT plan. Each thread has an access level, Read only, Edit files or Full access, with a Preview panel and a Changes view.
- Review. The agent records evidence against each criterion: test output, screenshots of each state and the diff. A person reviews it and decides the merge and release gates. An agent can request a gate but never approve one.
Outside Desktop, the same context is one MCP connection away. For Claude Code, run claude mcp add with the HTTP transport and the sydla.com/mcp URL. For Codex, add an mcp_servers.sydla table to config.toml with the same URL and your key in the headers. Cline and Kilo Code read remote MCP servers too, so an engineer who prefers Cline's Plan and Act loop or a Kilo custom agent can point it at the same brief and states. Keep the first prompt short: read the work item, follow the criteria, record evidence. The brief to pull request walkthrough shows one feature end to end, and the context engineering guide explains why the context should not live inside any one agent.
Five checks before you standardise on an open-source agent
Open-source agents are a good fit for teams that like to tune. Before you roll one out to the whole team, check these five things.
- Which plan pays. List the plan each engineer already has. ChatGPT plans work in all four agents; Claude plans only in Claude Code. Anything else means a new API bill.
- Gateway fine print. At cost or no markup is not the same as free: Kilo adds 5% when you buy credits, and cloud agents add compute by the hour.
- Spend caps. With BYOK, set a monthly limit at the provider before the first long run. Sign in with ChatGPT stops at the plan limit, which is safer by default.
- One rules file. Cline, Kilo Code and Codex read AGENTS.md; Claude Code reads CLAUDE.md. Keep one shared file and import it, as our CLAUDE.md vs AGENTS.md guide explains.
- Vendor risk. Roo Code closed in May and Kilo changed owners in July. Keep tickets, criteria and states outside the agent, so switching agents costs a config change, not a migration.
When Cline or Kilo Code is the right choice
Be fair to both. Cline is the most flexible agent on this list: open source under Apache 2.0, any model including local ones, Plan and Act, checkpoints and a large MCP marketplace. Kilo Code adds a real team product at $15 a seat with shared keys and analytics, cloud agents and code reviews, and the widest model gateway. If your team wants open models such as GLM or Kimi, or runs models on its own hardware, both beat Claude Code and Codex.
They are a weaker fit if your engineers already pay for a Claude plan, because that plan does not cover runs in either tool. They are also weaker when the problem is not the agent at all but a brief that lives in five places. Our Claude Code vs built-in AI agents guide, the Google Antigravity comparison and the AI credit fatigue guide look at the same question from other sides.
Try it on your next ticket
Sydla is $19 per user per month with every feature included and no AI credits to buy. Your Claude Code or Codex plan does the coding, and the brief, the states and the evidence live with the ticket. See the pricing section, read how Sydla works, then start your workspace and keep your Claude Code and Codex accounts.
Frequently asked questions
Is Cline free?
Yes. Cline's VS Code extension, CLI and MCP Marketplace are free and open source under Apache 2.0. You pay for model use, either with your own API key, with Cline credits bought at cost, with a ChatGPT plan through Codex sign-in, or with ClinePass at $9.99 a month for curated open models. Enterprise features such as SSO and audit logs are custom priced.
How much does Kilo Code cost?
The Kilo Code extension and CLI are free for individuals. Kilo Teams costs $15 per user per month with a 14-day trial. Model use is billed separately through your own key or Kilo Gateway credits at provider rates plus a 5% processing fee, and Kilo Pass prepaid plans cost $19, $49 or $199 a month. Cloud agents add $0.60 or $1.20 per hour of compute.
Can I use my Claude Pro or Max subscription in Cline or Kilo Code?
Not in Kilo Code: its docs say Claude subscriptions cannot be used in third-party harnesses, and its Claude linked billing needs a Claude API organization billed per token. Anthropic's terms reserve Pro, Max and Ultra logins for Claude Code and its own apps. Cline lists a provider that drives your local Claude Code CLI; check Anthropic's current terms before relying on it. Claude Code itself is the supported way to use a Claude plan.
Can I use my ChatGPT subscription in Cline or Kilo Code?
Yes. Cline has offered OpenAI Codex sign-in since January 2026, and Kilo Code supports ChatGPT Plus and Pro sign-in in its extension, CLI, cloud agents and code reviews. Runs spend your plan's included Codex allowance and stop at the plan limit instead of billing more.
Is Cline or Kilo Code better than Claude Code?
It depends on what you pay for. Cline and Kilo Code win on model choice, open source and local models. Claude Code wins if you already pay for a Claude plan, because that plan only covers Claude Code. Either way, the agent works best when it reads one brief, one set of criteria and the designed states from one place. Sydla holds those for $19 per user and runs your own Claude Code or Codex with zero token markup.



