Description
Cody is Sourcegraph’s AI coding assistant, built on Sourcegraph’s own code intelligence platform so it can retrieve context from an organization’s entire remote codebase, not just files open locally. Cody’s code graph indexes symbol definitions, references, and usages across hundreds of repositories, feeding that context into an admin-selected LLM for chat, autocomplete, and code edits. As of July 2025, Sourcegraph discontinued the Cody Free and Cody Pro plans that individual developers previously used, repositioning Cody as an enterprise-only product; individual users were redirected to a newer, separate Sourcegraph product called Amp. Cody Enterprise now requires an annual contract and includes admin-chosen model selection across Claude, GPT, and Gemini, a guaranteed no-training policy, SOC 2 compliance, and self-hosted or BYOC deployment for regulated environments. It suits large engineering organizations with sprawling, multi-repository codebases who need AI assistance that understands cross-repo architecture, not individual developers or small teams looking for a self-serve plan.
Key Features
- Cross-repository code graph — retrieves context from an entire organization’s codebase, across hundreds of repos.
- Admin-selectable LLMs — administrators choose which models (Claude, GPT, Gemini) power Cody for their organization.
- AI chat and autocomplete — explains code, generates functions, writes documentation and tests, and fixes bugs inline.
- Context Filters — fine-grained control over which code can be sent to LLMs, for compliance needs.
- No-training guarantee — customer code is never used to train AI models, per Sourcegraph’s policy.
- Self-hosted or BYOC deployment — deploy on Sourcegraph’s cloud or an organization’s own infrastructure for air-gapped environments.
- Broad IDE support — extensions for VS Code, JetBrains IDEs, and Neovim.
How It Works
Cody connects to an organization’s Sourcegraph Enterprise instance, which indexes the organization’s repositories using Sourcegraph’s existing code search and code intelligence infrastructure. Installing the Cody extension in VS Code, a JetBrains IDE, or Neovim connects it to that org’s endpoint, so questions, chat, and autocomplete requests draw on repository-wide context rather than only the currently open file. When a developer asks Cody a question or requests a code change, it retrieves relevant symbol definitions, references, and code from across the indexed repositories and feeds that context to whichever LLM the organization’s admin has configured. Context Filters let administrators restrict which code is eligible to be sent to an LLM at all, and every request is covered by Sourcegraph’s no-training guarantee on customer code.
Technical Architecture & Overview
- Core Engine: Not one model; administrators select from frontier models including Claude, GPT, and Gemini, layered on top of Sourcegraph’s code graph for context retrieval.
- Deployment: Requires a Sourcegraph Enterprise instance, either cloud-hosted or self-hosted/BYOC for air-gapped environments; extensions install for VS Code, JetBrains IDEs, and Neovim.
- API Surface: Integrates with an organization’s existing Sourcegraph code search deployment rather than a standalone public API.
- Known Limits: No individual or small-team self-serve plan exists as of this writing; Cody Free and Cody Pro were discontinued for new signups on June 25, 2025, and existing accounts were sunset by July 23, 2025.
Pros & Cons
| Pros | Cons |
|---|---|
| Cross-repository context is a genuine differentiator for organizations with hundreds of repositories | Free and Pro tiers were discontinued in mid-2025, shutting out individual developers and small teams |
| Admin-selectable model choice avoids lock-in to a single AI provider | Enterprise-only pricing requires an annual contract, not a self-serve monthly signup |
| No-training guarantee, SOC 2 compliance, and self-hosted deployment suit regulated industries | At roughly $59/user/month, Cody costs meaningfully more than GitHub Copilot Enterprise or Cursor Business |
| Sourcegraph’s underlying code search platform is mature, not a bolt-on AI feature | Only makes economic sense at larger team sizes, where the fixed enterprise overhead amortizes |
| Broad IDE coverage across VS Code, JetBrains IDEs, and Neovim | Individual developers are pointed to a separate product, Amp, rather than a lighter Cody tier |
Pricing
| Plan | Price |
|---|---|
| Cody Enterprise | Reported at approximately $59/user/month, annual contract (Cody Free and Pro discontinued mid-2025) |
Platform Availability
VS Code | JetBrains IDEs | Neovim | Web (via Sourcegraph)
Best For
Large engineering organizations | Multi-repository codebases | Regulated industries
Frequently Asked Questions
Is Cody still free to use?
No. Sourcegraph discontinued Cody Free and Cody Pro for new signups on June 25, 2025, and sunset existing accounts by July 23, 2025. Cody is now an Enterprise-only product with no free tier.
What happened to individual Cody users?
Sourcegraph redirected individual developers to a separate, newer product called Amp, positioning Cody exclusively for enterprise engineering organizations going forward.
How much does Cody Enterprise cost?
Multiple independent sources report Cody Enterprise at approximately $59 per user per month on an annual contract, though exact enterprise pricing can vary by negotiated deal and should be confirmed directly with Sourcegraph sales.
What makes Cody different from GitHub Copilot or Cursor?
Cody retrieves context from an organization’s entire remote codebase across many repositories using Sourcegraph’s code graph, while Copilot and Cursor rely primarily on local file context within a single repository.
Which IDEs does Cody support?
Cody has official extensions for VS Code, JetBrains IDEs (including IntelliJ, PyCharm, and WebStorm), and Neovim, connecting to an organization’s Sourcegraph Enterprise instance.
Similar Tools in This Directory
Also in AI No-Code Builders: Sweep, next-edit autocomplete for JetBrains | Aider, open-source AI pair programmer | Bloop, AI code search | Eidolon AI, open-source agent server
