Continue.dev was an Apache 2.0 AI coding assistant for VS Code and JetBrains that let you plug in any model, including local ones through Ollama. Cursor's parent company Anysphere acquired it around June 16, 2026 in what amounts to an acqui-hire: a final 2.0.0 release shipped, the GitHub repository went read-only, recurring billing was switched off, and hosted data was deleted after July 15, 2026. The extension you already have installed still runs. Nobody is fixing it, and nobody should adopt it now.
Reviewed by EdGrows·Updated Aug 18, 2026Affiliate link — we may earn a commission
This analysis is based on documentation, public user reports, and vendor materials — not yet on our own hands-on testing. How we rate
Around June 16, 2026, Continue's homepage quietly changed. No press release, no launch post — an FAQ explaining that Cursor's parent company Anysphere had acquired the team, and what current users needed to do about it.
The sequence since:
Terms weren't disclosed. The shape of it is an acqui-hire: Cursor wanted the people and the accumulated knowledge of building agent UX, and the standalone product wasn't part of the plan.
Context worth having: the deal was signed in the same window that SpaceX agreed to acquire Anysphere itself for $60 billion in an all-stock transaction. Continue's code now sits inside Cursor, which sits inside something considerably larger.
Nothing broke on July 15. That's what makes this awkward.
Version 2.0.0 still installs, still talks to your API keys, still runs local models through Ollama. There's no error message, no shutdown notice in the editor, no forced migration. For a while it will feel like the tool you chose.
The decay is in what stops happening. New model endpoints won't be integrated as providers ship them. Bugs won't be fixed. Security issues won't be patched. And when VS Code or JetBrains change an extension API — which they do routinely — nobody is on the other end to adapt.
That gives you time rather than urgency. Use it. Migrating on your own schedule beats migrating the week something stops loading.
The code is Apache 2.0, which is the one genuinely reassuring detail here. Teams who forked, vendored, or built internal tooling on Continue's architecture keep the right to carry on.
Whether they should is a resourcing question rather than a licensing one. A fork means inheriting model-provider churn, IDE compatibility work, and security review — permanently. Organisations with platform engineering to spare can absorb that. Individuals almost never can, and the honest advice for a solo developer is to move rather than maintain.
There's a broader lesson in the licence, though, and it's worth stating plainly: users of proprietary tools that got acquired in 2026 had no equivalent option. When Codeium became Windsurf became Devin Desktop, nobody could fork it. Continue's users at least got a copy of the thing they depended on.
Pick by the reason you chose Continue originally, since it appealed to several different people for different reasons.
You were there for JetBrains support. JetBrains AI Assistant is native, starts at $10 a month, and understands the IDE's semantic model in ways an extension can't. That's the closest thing to a straight upgrade in this list.
You were there for model freedom and BYOK. Cline inside VS Code or Aider in the terminal. Both open source, both actively maintained, both keep the "any model, my key" property intact.
You were there for local-only inference. Cline with Ollama, or Aider pointed at a local endpoint. Neither sends code anywhere you don't configure.
You just wanted free AI help in VS Code. GitHub Copilot has a free tier and unmetered completions on paid plans. Least effort, biggest company behind it.
Against Cline: Cline is the natural successor for most Continue users — open source, bring your own key, agentic, and still shipping. It's VS Code-centric, so JetBrains users lose something in the move. Everything else transfers, including the philosophy.
Against Aider: Aider covers the same model-freedom ground from the terminal instead of the editor, with deeper git integration and better token efficiency. The adjustment is ergonomic rather than conceptual.
Against JetBrains AI Assistant: JetBrains AI Assistant is the answer for the JetBrains half of Continue's audience. It's paid and credit-metered rather than free and BYOK, but it's native, supported, and not going anywhere.
Against Cursor: Cursor is where Continue's team went, and it's the migration path the acquisition implicitly suggests. It's a full editor at $20 rather than an extension in yours, and it's proprietary — so you're trading the property that made Continue appealing for polish and momentum.
| Plan | Status |
|---|---|
| Open-source extension | Free, Apache 2.0, v2.0.0 final, unmaintained |
| Solo (hosted) | No longer sold |
| Teams (listed at $20/seat with $10 model credits) | No longer sold; billing disabled |
| Enterprise | No longer sold |
Checked August 2026. These plans are a record, not an offer — the hosted platform wound down following the June 2026 acquisition and recurring billing was disabled. Reported pricing varied across sources during the wind-down (Teams was listed at $20 per seat in some documentation and $10 in others), which is itself a reason not to rely on any current figure. The extension remains free under Apache 2.0.
Export anything you still have locally, then treat 2.0.0 as a stable snapshot with an expiry date you get to choose.
Continue deserved better than a homepage FAQ. It was one of the earliest open-source coding assistants, it supported JetBrains when almost nothing else credible did, and its bring-your-own-model design was the correct architecture — the thing plenty of developers wanted and few vendors offered. Over 2.5 million VS Code installs and 33,000-plus GitHub stars by May 2026 say the demand was real.
None of that saved it. The repository is read-only, the hosted platform is gone, and the team works at Cursor now — which is itself being folded into SpaceX. The extension you have installed keeps running, and that's the trap: there's no failure event to force a decision, just slow drift as models, IDEs and security move on without it.
The Apache 2.0 licence is the one thing that separates this from the other 2026 disappearances. You can fork it, and some teams should. Most shouldn't, because inheriting perpetual maintenance to avoid a one-week migration is a bad trade.
Not recommended for adoption under any circumstances. If you're running it, migrate deliberately over the next few months rather than in a hurry later.
Note: Continue does not have an affiliate program with AIVario. We earn no commission, and this rating reflects the product's discontinued status rather than the quality of the work.
Best for: Existing users planning a migration; teams thinking seriously about fork risk before their next tool choice Not ideal for: Anyone choosing a coding assistant today Bottom line: Good architecture, real adoption, acquired and frozen in June 2026. The code survives under Apache 2.0; the product doesn't.
The product is, the code isn't. Cursor's parent company Anysphere acquired Continue around June 16, 2026, announced through a homepage and FAQ update rather than a press release. A final 2.0.0 release shipped covering the VS Code extension, JetBrains plugin and CLI, after which the continuedev/continue repository went read-only — no new issues, no merged pull requests, no roadmap. The Apache 2.0 code remains public and forkable.
Deleted after July 15, 2026, if you didn't export it. Hosted conversation history, saved configurations and team settings were all subject to that deadline. Recurring billing was disabled at the same time, so existing subscribers were not charged again. If you missed the window, the local extension configuration on your own machine is unaffected — it's the cloud-side material that's gone.
Yes, with caveats that grow over time. Version 2.0.0 still installs and runs against your own API keys or local models. What stops is everything downstream: no new model integrations as providers ship new endpoints, no bug fixes, no security patches, and no compatibility work when VS Code or JetBrains change something underneath it. It degrades rather than breaks, which is the more annoying failure mode because there's no clear moment to act.
Legally yes, practically it depends on your appetite. Apache 2.0 permits forking, vendoring and internal redistribution, and teams who built tooling on Continue's architecture can keep going. What a fork doesn't give you is the maintenance burden disappearing — you inherit model-provider churn, IDE API changes, and security review. For an organisation with the engineering to spare it's viable; for an individual it's usually more work than switching tools.
For the team and the pattern library, not the product. Continue was one of the earliest widely adopted open-source coding agents, with agent UX patterns, model orchestration experience and a large contributor network. Cursor has been consolidating aggressively — and is itself being absorbed, with SpaceX agreeing on June 16, 2026 to acquire Anysphere for $60 billion in stock. Continue's code now sits inside a company that sits inside a much larger one.
Match the reason you chose Continue in the first place. If it was JetBrains support, JetBrains AI Assistant is native and starts at $10. If it was model freedom and BYOK, Cline or Aider both keep that intact and are actively maintained. If it was local-only inference for privacy, Cline with Ollama covers it. If it was simply free AI assistance in VS Code, GitHub Copilot's free tier is the least effort.
Still the fastest issue tracker for engineering teams — and after the plan restructure, the tier with the good features costs $16 instead of $50.
The AI code editor that became the default — now rebuilt around parallel agents in Cursor 3, with its own Composer models.
Terminal-native, model-agnostic, git-obsessed. Cheaper than the vendor CLIs and slightly less accurate — an explicit trade, not a hidden one.