Sourcegraph Cody

No. Sourcegraph discontinued Cody Free and Cody Pro in July 2025, and new sign-ups for those plans stopped on June 25, 2025. Today Cody is only available as part of Sourcegraph Enterprise, which starts at a $16,000 minimum annual contract and scales with team size; the plan includes AI credits per user, but Sourcegraph publishes no per-seat price. If you are an individual looking for a free alternative, consider Continue.dev, Cline, or Aider, all of which are open source.

The differentiator is context. Cody uses Sourcegraph's code graph to pull context from your team's entire remote codebase at any scale, not just files you happen to have open locally. For large enterprise monorepos with millions of lines of code, this is a meaningful advantage over GitHub Copilot and Cursor, which rely primarily on local file context. If your company already runs Sourcegraph for code search, Cody is a natural extension.

Cody is sold through Sourcegraph Enterprise, which starts at a $16,000 minimum annual contract and scales with team size. The plan includes credits for AI features per user, pooled across the org, but Sourcegraph publishes no per-seat price, so you will need a quote to compare it against GitHub Copilot Enterprise or Cursor Business. The $16,000 is the price of the Sourcegraph Enterprise platform; the pricing page does not list Cody separately. The premium is justified if your team needs deep remote codebase context, enterprise security controls, and the Sourcegraph platform for code search alongside Cody.

Yes, in most setups. Cody has official extensions for VS Code and JetBrains IDEs (IntelliJ IDEA, PyCharm, WebStorm, GoLand, PhpStorm, Rider, RubyMine, CLion, Android Studio and more), plus an experimental extension for Visual Studio. You can also use Cody in the Sourcegraph web app and through the Cody CLI, which is experimental. A Neovim plugin (sg.nvim) still exists, but Sourcegraph says it is experimental and no longer actively maintained. Features include chat, code edits, autocomplete, and reusable prompts.

No. For Enterprise customers, Sourcegraph commits to not training on your company's data, and its third-party LLM providers do not train on your codebase. To answer a question, Sourcegraph retrieves relevant code snippets that the user has permission to read and sends them to the LLM provider as context. This is a key reason regulated industries and enterprise security teams pick Cody over free consumer tools that may have opaquer data policies.

Cody Enterprise lets admins configure which LLMs are available. Sourcegraph's supported-models list includes Claude Opus 5, Claude Sonnet 5 (the default chat model), Claude Haiku 4.5, Gemini 3.8 Flash and other Gemini 3.x Flash models, Gemini 3.1 Pro (beta), and GPT-5.6. Enterprise customers can also bring their own API keys (BYOK). This flexibility is useful for regulated customers that need specific provider agreements.

Cody is more focused on chat, code edits, autocomplete, and prompts than on fully autonomous agentic workflows, though it supports the Model Context Protocol (MCP) for connecting external context. For agentic coding, Sourcegraph built a separate tool, Amp, and its platform now also offers Agentic Batch Changes for code changes at scale. If your primary use case is long-running autonomous agents, Cursor Composer, Cline, or Claude Code are still ahead. Cody's strength is deep context over huge codebases.

For large enterprises with remote monorepos, yes. Cody's full-codebase context, model flexibility, and Sourcegraph search integration are unique and can justify Sourcegraph Enterprise's platform pricing, which starts at $16,000. For teams that already live in GitHub and want the cheapest enterprise AI option, Copilot Enterprise may be a better fit. Evaluate both in a 30-day pilot on your real codebase before committing.