Continue
The free, open-source AI code assistant that lives in your IDE — and lets you bring any model, including local ones at $0.
About
Continue is the free, open-source AI code assistant that lives inside your editor — and the most direct answer to a swollen GitHub Copilot bill. When Copilot moved to usage-based, per-token billing on June 1, 2026, developers watched flat $10 plans turn into metered invoices (one report jumped from $29 to nearly $750 a month). Continue's pitch is the opposite: the tool is free, and you bring your own model. It installs as an extension for both VS Code and JetBrains (IntelliJ, PyCharm, WebStorm and the rest), giving you Copilot-style chat, tab autocomplete, and inline edits. The difference is who you pay. Continue connects to whatever you choose — Claude, GPT, Gemini, or a local model running through Ollama, LM Studio or Jan. Run a local model and your per-token cost is exactly $0, with your code never leaving your machine. You can even mix providers: a fast local model for completions, a frontier cloud model for chat. That flexibility is the trade. Because you bring your own model, there's setup — you configure providers and API keys yourself rather than clicking sign-in. Local models also lean on your hardware; weak machines won't match cloud quality, and the autocomplete isn't quite as deeply wired as Copilot's native tab experience. The project has also been growing into a hosted CI 'AI checks' product, so the open-source assistant is now one piece of a broader offering. Verdict: if you're a solo dev or small team leaving metered Copilot, Continue is the easiest in-IDE landing spot — free, model-agnostic, and private by default. It's the practical 'stop paying the meter' move for editor users. Read our breakdown of why Copilot's new billing pushed devs to free tools, and if you'd rather work in the terminal, pair it with Aider.
Key Features
- Free, open-source extensions for both VS Code and JetBrains (IntelliJ, PyCharm, WebStorm and more)
- Bring your own model: Claude, GPT, Gemini, local models via Ollama, or any OpenAI-compatible API through OpenRouter
- Copilot-style chat, tab autocomplete, and inline edits inside the editor
- Mix and match — use a local model for completions and a cloud model for chat
- Connect to Ollama, LM Studio, Jan, or paid APIs like OpenAI and Anthropic with a single config file
- No vendor lock-in and no subscription required to use the core assistant
Use Cases
- 1Replacing a metered GitHub Copilot subscription with a free extension that runs on your own API key
- 2Running a fully local coding assistant via Ollama with zero cloud dependency and zero per-token cost
- 3Keeping code private — point Continue at a local model so nothing leaves your machine
- 4Standardizing one assistant across a mixed VS Code and JetBrains team without per-seat lock-in
Pros
- Genuinely free and open-source (Apache-2.0) — no per-token meter on the tool itself, unlike Copilot's June 2026 usage-based billing
- Model choice is the whole point: 75+ provider options means you're never locked to one vendor's price or quality ceiling
- Runs local models via Ollama/LM Studio/Jan for literally $0 per token and full privacy
- Works in both VS Code and JetBrains with the same core features (chat, autocomplete, inline edit)
Cons
- BYO model means setup work — you configure providers and keys yourself instead of one-click sign-in
- Local models need decent hardware (VRAM) to match cloud-model quality; weak machines feel the gap
- Autocomplete polish isn't quite Copilot's deeply-wired tab experience, especially on smaller local models
- The project has been expanding into hosted CI 'AI checks,' so the open-source assistant is one part of a broader, evolving product
Get Started
Details
- Category
- code
- Pricing
- Free