Continue.dev Alternatives After Cursor Bought Continue

The best Continue.dev alternatives depend on which part of Continue you relied on. For an open-source, bring-your-own-model agent in VS Code or JetBrains, move to Cline or Kilo Code. For local-model autocomplete, look at Tabby or Zed. For the Continue CLI, OpenCode is the closest fit. If you'd rather stay inside a big vendor's tool, JetBrains AI Assistant and GitHub Copilot now accept your own models for chat. And if you want to follow the Continue team, that road leads to Cursor.

You need a plan because Continue itself has stopped. Its homepage now reads "Continue was acquired by Cursor," and the GitHub repository is read-only. The extension you installed still exists, but nobody is fixing it.

Here's what that means in practice:

• No more updates. The last release went out in June 2026.
• The code is still yours to use. It stays Apache 2.0, so forks are allowed.
• Your setup carries over in pieces. Your model list, API keys and rules files can move to most of the tools below.

Below: what happened, a quick comparison, then the best replacement for each way people used Continue.

What Happened to Continue

Continue's homepage is now a thank-you page titled "Continue has joined Cursor." It says the team's open-source codebase "remains freely available as a foundation for others." The page doesn't give a date or say what Cursor will do with the product.

The Continue GitHub repository fills in the rest. The README now opens with a note that the repo "is no longer actively maintained and is read-only for all users." Before stopping, the team shipped a final 2.0.0 release of the VS Code extension, the CLI and the JetBrains plugin. That release removed anonymous telemetry and pulled out authentication. The GitHub releases list dates it to June 19, 2026.

Two details matter if you're deciding how fast to move:

So Continue keeps working until something around it breaks: a VS Code API change, a model provider changing its endpoints, a security issue. For a personal setup, that might be fine for a while. For a team, it's a reason to switch now while you can do it on your own schedule.

Quick Comparison

If You Used Continue as an Open-Source Agent

Most people who installed Continue in the last year used it as a chat and agent panel inside their editor, pointed at whatever model they liked. These two keep that shape.

Cline: the closest all-round replacement

Cline is an open-source coding agent that started in VS Code and now runs much wider: VS Code, Cursor, Windsurf and JetBrains extensions, a CLI, a desktop app and an SDK. Its homepage puts the pitch plainly: "Bring your own key or your own weights." That covers Claude, GPT and Gemini, local models through Ollama or LM Studio, and any OpenAI-compatible endpoint.

The pricing page says Cline "is free for individual developers," with no seat fees for the open-source version. You pay your model provider for usage.

Pick it if you want the biggest community and the most surfaces, with the same "my model, my keys" freedom Continue gave you. The trade-off: Cline is agent-first. It plans, edits and runs commands, and on large tasks that can burn tokens faster than Continue's chat ever did. Our Cline alternatives guide covers its weak spots.

Kilo Code: one agent everywhere you work

Kilo Code is MIT-licensed and runs in VS Code (and Cursor, Windsurf and other Open VSX editors), JetBrains IDEs, a CLI, a browser-based cloud agent, Slack and a mobile app. You can bring your own keys or run local models, and Kilo says it adds zero markup on AI inference when you use its hosted models.

Pick it if your team uses more than one editor and wants a single open-source agent with the same settings across all of them. It's a good fit for ex-Continue JetBrains users, since the JetBrains extension is a first-class surface rather than an afterthought.

Roo Code, the third big agent in this family, shut down in May 2026. If you're comparing Cline and Kilo Code closely, our Roo Code alternatives post goes deeper on how they differ.

If You Used Continue for Local Autocomplete

A big reason people chose Continue was tab completion from a model running on their own machine. That niche is smaller now, but it isn't empty.

Tabby: self-hosted completion for a team

Tabby is an open-source, self-hosted coding assistant. You run the server yourself, with no external database or cloud service needed, and connect editors to it. It offers code completion, inline chat and an answer engine, with plugins for VS Code, Neovim, Eclipse and the JetBrains family.

Pick it if you want one internal completion server that a whole team shares, and code must never leave your network. The honest trade-off: activity has slowed. The Tabby repository is still public and not archived, but its last tagged release was v0.32.0 in January 2026, and TabbyML's homepage now leads with a newer cloud agent called Pochi. Check the repo before you commit a team to it. If you need a vendor-backed private deployment instead, our Tabnine alternatives guide covers that market.

Zed: switch editors and get local AI built in

Zed is an open-source code editor with AI built in rather than bolted on. Its Zeta model predicts your next edit, the agent panel handles multi-file changes, and you can "connect models through Ollama and keep everything on your hardware." Zed can also host outside agents such as Claude Agent, Codex and OpenCode.

Its pricing page lists a free Personal plan for use with your own API keys or external agents, and paid plans if you want Zed's hosted models.

Pick it if you're open to leaving VS Code or JetBrains. You get fast local edit predictions without stitching an extension and a model server together. If changing editors is a dealbreaker, Tabby or Cline is the better route.

If You Used the Continue CLI

Continue's final release pushed people toward its CLI, especially JetBrains users. If you liked running an agent from the terminal or in CI, a terminal-first tool is the natural next step.

OpenCode: the open-source terminal agent

OpenCode is an MIT-licensed coding agent with a terminal UI that "works in any shell, editor or remote server," plus a desktop app and an SDK for scripts and CI jobs. It supports 200+ model providers, including local models, and ships with a set of free models that need no account. You can also sign in with an existing ChatGPT or GitHub Copilot subscription.

Pick it if you want an actively maintained, model-agnostic CLI with a large community. The trade-off is the same as any terminal agent: no inline tab completion, so pair it with an editor tool if you need that.

If You'd Rather Stay With a Big Vendor

Some teams picked Continue because they couldn't get approval for a closed tool, or because they needed local models. Both big vendors now cover part of that ground.

JetBrains AI Assistant

If you're all-in on IntelliJ, PyCharm or WebStorm, JetBrains AI Assistant lets you plug in API keys for Anthropic, Google, OpenAI or any OpenAI-compatible endpoint. It also runs local models through Ollama and LM Studio.

Read the limits, though. JetBrains' docs say "Agents cannot use local models," and local models can't call MCP tools. Some features may also need a JetBrains AI subscription when you use third-party models. Pick it if you want native JetBrains integration and mostly need chat with local or custom models.

GitHub Copilot with your own models

GitHub Copilot in VS Code now has a bring-your-own-key option for chat, and VS Code's docs say local models like Ollama work "without a Copilot plan and without signing into a GitHub account." The catch is in the same doc: you "cannot connect to a local model for inline suggestions." Business and Enterprise users also need an admin to enable the policy.

Pick it if your company already pays for Copilot and you want chat on a local or custom model without adding another tool. For the wider field, see our GitHub Copilot alternatives.

What About Cursor?

The Continue team went to Cursor, so it's fair to ask whether you should too. Cursor's own blog says Cursor was acquired by SpaceX in August 2026, and its plans now sit alongside SpaceXAI's Grok models, with third-party models billed separately by usage.

Cursor is a strong editor with deep agent features. But it's the opposite of what most Continue users wanted: a closed-source editor of its own, with plans built around hosted models. Pick it if you want the most polished AI editor and don't care about open source. Skip it if privacy or model choice was why you chose Continue.

How to Choose Your Continue Replacement

Start with what you actually used last month:

Before you switch:

1. Save your config. Copy your Continue model list, rules and prompt files somewhere outside the extension folder. Rules usually move over with light edits.
2. Test with your real models. If you ran a specific local model, confirm the new tool handles it well. Tool-calling support varies a lot between local models.
3. Decide on forking honestly. You can fork Continue under Apache 2.0, but then you own every future fix. For most teams, moving is cheaper than maintaining a fork.

Where Taku Fits

Taku isn't a coding assistant, so it doesn't replace Continue. It solves a different problem: you've seen someone share a working AI setup, such as a local-model coding workflow or a research agent, and you can't get it running on your own machine.

Taku is an AI-native desktop workspace where you can mirror a working AI app or workflow someone shared, remix it, and keep it instead of re-prompting from scratch. Taku is in Beta, and the Mac app is available now.

FAQ

Is Continue.dev shut down?

Development has stopped. Continue's homepage says it was acquired by Cursor, and its GitHub repo says it is no longer actively maintained and is read-only. The final 2.0.0 release can still be installed, but it won't get fixes.

What is the best free Continue.dev alternative?

Cline is the closest free swap. It's open source, free for individual developers, and works with your own API keys or local models. Kilo Code and OpenCode are also open source, and Zed's Personal plan is free with your own keys.

Which Continue alternative works best with local models?

For an agent, Cline and Kilo Code both support local models. For tab completion, Tabby runs a self-hosted server and Zed offers local edit predictions in its editor. GitHub Copilot and JetBrains AI Assistant support local models for chat only.

Is there a Continue alternative for JetBrains?

Yes. Kilo Code and Cline both have JetBrains extensions, Tabby has JetBrains plugins, and JetBrains AI Assistant works with your own keys and local models for chat. Continue's own README recommended its CLI over the JetBrains plugin before it stopped.

Did Cursor replace Continue?

Not as a product. Cursor acquired the company, but Continue's pages don't describe any merge of features. Cursor remains a separate, closed-source editor.

Can I keep using Continue?

You can, at your own risk. The code is Apache 2.0 and the final release still installs. Over time, editor updates and model API changes can break it, and no one will patch it unless you or a community fork does.