GitHub Copilot Alternatives: Pick by Why You're Leaving

The best GitHub Copilot alternatives depend on why you're leaving. If Copilot's new AI Credits make your bill unpredictable, Claude Code or Codex on a flat subscription is the fix. If your code can't leave your machine, OpenCode or Cline with a local model is. If you want a cheaper start, Codex is included even in ChatGPT's free plan.

Most people searching for alternatives in 2026 have a specific complaint. GitHub moved Copilot to usage-based billing on June 1, 2026 and paused new sign-ups for months. So this guide is sorted by reason, not by a single ranking.

How we researched this: official docs, pricing pages and changelogs as of October 2026, plus public user reports where linked. We haven't run our own benchmarks.

GitHub Copilot alternatives compared

Prices are as of October 2026, from each vendor's own pricing page.

For reference, Copilot Pro costs $10 a month and includes $15 of model usage plus unlimited completions. Our GitHub Copilot review covers what changed and who should stay.

If you're leaving because of unpredictable limits

Pick Claude Code. Copilot now charges chat and agent work by tokens, and part of each paid allowance is a "flex allotment" that GitHub says can change. Claude Code comes with every paid Claude plan, per Claude's pricing page: Pro is $20 a month ($17 billed annually), and Max starts at $100.

Claude's limits reset on a rolling five-hour window, with weekly limits on top for paid plans, rather than one monthly token pool. You still have limits, but a heavy Monday won't eat your whole month. Claude Code runs in the terminal, VS Code, JetBrains IDEs, a desktop app and the browser, per its docs. Our Claude Code pricing guide breaks down what each plan gets.

The downside: there's no free tier for Claude Code. Claude Free doesn't include it.

If you're leaving because of price

Pick Codex. OpenAI's Codex pricing page says Codex is included in ChatGPT Free, Go, Plus, Pro, Business, Edu and Enterprise. Go costs $8 a month and Plus $20. Free users get Codex with GPT-6 Luna in the desktop app, subject to rollout.

OpenAI doesn't publish numeric limits for Free and Go. For Plus, it estimates 350 to 3,000 local messages per five hours on GPT-6 Luna, and far fewer on bigger models. Pro plans, from $100 a month, currently have no five-hour limit.

If you pay for ChatGPT anyway, Codex costs you nothing extra. See how it stacks up in our Claude Code vs Codex comparison.

If you want a stronger or more flexible agent

Copilot's cloud agent is useful, but it has firm limits: 59 minutes per session, one repository per task, and GitHub-hosted repos only, per GitHub's cloud agent docs. If those get in your way, you have three good options.

• Claude Code for the strongest single-vendor agent, with terminal, IDE, desktop and web versions.
• Codex for an agent bundled with your ChatGPT plan, available in the CLI, an IDE extension, the desktop app and Codex Cloud.
• OpenCode if you want an open-source agent that isn't tied to one model company.

OpenCode is MIT-licensed and connects to 75+ providers through Models.dev, including local models. It also ships some free models that work without an account. Its model list shows which ones may use your data for training during the free period, so check before you paste private code.

If you need privacy or self-hosting

Pick OpenCode or Cline with a local model through Ollama or LM Studio. Nothing leaves your machine, and the only cost is your hardware.

Cline is a free, open-source VS Code extension and CLI. You pay only for model usage, or nothing at all if you point it at a local model. Its Ollama setup docs suggest 32 to 64 GB of RAM for mid-size coding models.

For the model itself, Qwen3-Coder-Next on Ollama is an 80B mixture-of-experts model with 3B active parameters and a 256K context window. The default download is 52 GB, so this is a workstation setup, not a laptop one. The smaller Qwen3-Coder 30B is 19 GB.

If your company needs a self-hosted server rather than a desktop setup, Tabby bills itself as an open-source, on-premises Copilot alternative with VS Code, Vim and IntelliJ plugins. Note that the latest Tabby release its README names is v0.30, from 2025, so check the project is still active before you build on it.

On Copilot's side, GitHub says it doesn't train on Business or Enterprise data. Free, Pro and Pro+ users are opted in to training on interaction data unless they turn it off, per the Copilot features page.

If you use JetBrains or another editor

Copilot does support JetBrains IDEs, but some features land in VS Code first. The 1 million token context window, for example, is VS Code and CLI only.

JetBrains users should look at JetBrains AI and its Junie agent. The Junie page lists a free Junie Lite model in every JetBrains account, AI Pro at $8.33 a month with 10 AI credits, and AI Ultimate at $25 a month with 35 credits. Those monthly figures match the $100 and $300 yearly prices. You can also bring your own key at provider rates with no markup, including local models.

The catch: JetBrains' own licensing docs show two different price tables, and user posts on JetBrains' support forum complain that credits run out quickly. Check the price at checkout.

If you'd rather switch editors entirely, you have two clear choices:

• Cursor is an AI-first editor. As of October 2026, Cursor's pricing is built around Grok: Plus at $8 a month only gets Grok models, and Claude, GPT and Gemini need Premium at $30, billed separately at API list price. Our Cursor vs Claude Code comparison goes deeper.
• Zed is a fast, open-source editor. Its Personal plan is free with your own API keys or external agents like Claude Agent and Codex CLI. Pro is $10 a month with unlimited edit predictions and $5 of tokens. Zed even supports GitHub Copilot as a provider, so you can keep Copilot and change only the editor.

Other options worth a look

Kiro gives free users 50 credits a month with Claude Sonnet 4.5 and open-weight models like Qwen3 Coder Next. Pro costs $20 for 1,000 credits, and extra credits are $0.04 each.

Google's Antigravity has a free individual tier with unlimited tab completions and a weekly rate limit. It replaced Gemini Code Assist for individuals and the free Gemini CLI login, which stopped working on June 18, 2026.

Windsurf's old pricing URL now redirects to Devin's pricing page, which lists a free plan with a light agent quota and unlimited tab completions, Pro at $20 and Max at $200.

Which Copilot alternative should you choose?

• Choose Claude Code if you want the most capable agent and a plan that resets every few hours instead of once a month.
• Choose Codex if you already pay for ChatGPT, or want a free agent to start with.
• Choose OpenCode or Cline with a local model if code can't leave your machine.
• Choose JetBrains AI if your team works in IntelliJ, PyCharm or WebStorm.
• Choose Zed if you want a lighter editor and your own API keys.

Staying on Copilot still makes sense if you mostly use completions and light chat. At $10 a month with unlimited completions, it's hard to beat on value.

If cost is the main reason you're looking, our list of the best free AI for coding shows the exact free limits of each tool.

FAQ

What is the best free alternative to GitHub Copilot?

Codex through ChatGPT Free is the easiest free agent, and Cline or OpenCode with a local model is the only option with no usage cap. Copilot Free itself still includes 2,000 completions a month.

Is Cursor better than GitHub Copilot?

Cursor is an AI-first editor, while Copilot adds AI to the editor you already use. Since October 2026, Cursor's cheapest paid plan only includes Grok models, so Copilot Pro is better value if you want Claude or GPT models for $10.

Can I use GitHub Copilot alternatives in JetBrains IDEs?

Yes. JetBrains AI and Junie are built in, Claude Code has a JetBrains plugin, and Cline offers a JetBrains extension on its Enterprise plan.