Cody AI Review 2026: Pricing, Features and Verdict
This review is researched from each provider's official pricing, plans and public user feedback — see our editorial process for how we keep it accurate.
Is Cody AI still worth using in 2026?
Cody is Sourcegraph's AI coding assistant, built on the company's code-search engine so it can pull context from an entire codebase rather than just the open file. The catch: Sourcegraph sunset the self-serve Free and Pro plans on July 23, 2025, so Cody is now bundled only into Sourcegraph Enterprise — a sales-led, five-figure-minimum contract, not a $10/month IDE plugin.
At a glance
| Cody AI | |
|---|---|
| Starting price | Not self-serve — part of Sourcegraph Enterprise, minimum annual contract around $16,000 |
| Free tier | None as of July 23, 2025 (Free/Pro/Enterprise Starter plans were sunset) |
| Best for | Enterprises already on Sourcegraph who need AI chat/autocomplete tied to large, multi-repo codebase search |
| Standout feature | Context pulled from Sourcegraph's code-search index across an org's entire codebase, not just open files |
What Cody actually is
Sourcegraph built its name on enterprise code search — the ability to search and navigate across every repository a large organization owns, not just the one open in an editor. Cody is the AI layer on top of that index: a chat and autocomplete assistant that uses Sourcegraph's search API to retrieve relevant functions, symbols, and usage patterns from across an entire codebase before generating an answer or a completion.
That's the pitch that separates Cody from most AI coding assistants: instead of relying only on the handful of open tabs and files an editor can see, it can, in principle, reason about how a function is used three repositories away. Sourcegraph doesn't publish an exhaustive model list in its public docs, but its documentation names Anthropic's Claude models as the default backbone for chat and edits, with OpenAI models available as an alternate configuration, and a DeepSeek-Coder-V2 model (hosted via Fireworks.ai) used for autocomplete. Enterprise customers can also bring their own API keys or route to self-hosted models, which matters for regulated industries that can't send code to a third-party inference endpoint.
Cody ships as extensions for VS Code, JetBrains IDEs, Visual Studio, and a web app, plus a CLI for scripted or agentic use.
The pricing situation you need to know about first
This is the part that makes Cody different from every other assistant in this comparison set: it no longer has a self-serve plan. Sourcegraph's own FAQ states plainly that "starting July 23, 2025, Cody will no longer be available for Free, Pro and Enterprise Starter plans," and that the change "will not impact Cody Enterprise customers." Individual developers and small teams who want a Sourcegraph-branded AI assistant are being pointed instead toward Amp, Sourcegraph's newer, separately billed agentic coding product — Cody itself now lives exclusively inside Sourcegraph Enterprise.
Sourcegraph's public pricing page lists Enterprise at a minimum annual contract "starting at $16K," which includes a pool of credits for AI features that scales with team size, with volume credit buckets available as an add-on and no monthly credit expiry. There's no listed per-seat monthly price and no self-serve checkout — every Enterprise deal goes through a sales conversation.
| Plan | Price | What you get |
|---|---|---|
| Free | Discontinued July 23, 2025 | No longer available for new or existing self-serve users |
| Pro | Discontinued July 23, 2025 | Existing users were offered $10–$40 in migration credits toward Amp |
| Enterprise Starter | Discontinued July 23, 2025 | Folded into full Enterprise |
| Enterprise | From $16,000/year minimum, custom quote | Cody chat/autocomplete/edit, full Sourcegraph code search, Deep Search, Batch Changes, Insights, Monitoring, MCP server/API/CLI access, single-tenant deployment, enterprise security controls, 24×5 support |
Pricing, plan availability, and feature sets were accurate as of this post's publish date and can change quickly for AI coding tools — confirm current terms directly with Sourcegraph sales before budgeting around them.
Core features that still differentiate it (inside Enterprise)
- Codebase-wide context retrieval. Cody's chat and completions are grounded in Sourcegraph's search index, so answers can cite where a function is defined and every place it's called, across repos — a meaningfully different approach from assistants that only see open files or a local embeddings cache.
- Deep Search. Sourcegraph markets an AI-powered search mode built for large, sprawling codebases, aimed at answering architecture and dependency questions that a single-repo assistant can't.
- Batch Changes with agentic assistance. Sourcegraph's Batch Changes tooling — for making the same code change across hundreds of repositories at once — now has an AI layer to help draft and apply large-scale refactors, which is a genuinely enterprise-scale use case most consumer coding assistants don't attempt.
- Bring-your-own-model / self-hosted routing. Enterprise customers can point Cody at their own model deployment or API keys rather than relying solely on Sourcegraph's hosted inference, which matters for companies with strict data-residency or model-approval requirements.
- Multi-IDE support tied to one backend. VS Code, JetBrains, Visual Studio, and a web client all share the same context engine, so a team standardized on Cody isn't locked to a single editor.
Who it's actually for
- Large engineering orgs already paying for Sourcegraph. If your company already licenses Sourcegraph Enterprise for code search across dozens or hundreds of repos, adding Cody is a natural extension — the context engine is already there.
- Teams doing large-scale, cross-repo refactors. The Batch Changes + AI combination is aimed squarely at "update this pattern across 300 repositories" work that individual-developer tools aren't built for.
- Regulated or security-conscious enterprises. Self-hosted model routing and single-tenant deployment options matter for finance, healthcare, and government engineering teams that can't send code to arbitrary third-party APIs.
- Not a fit for solo developers, freelancers, or small startups. With no free tier and a five-figure minimum contract, Cody is now priced entirely out of reach for individuals and small teams — those users are the exact audience Sourcegraph is steering toward Amp instead.
Pros and cons
| Pros | Cons |
|---|---|
| Genuine cross-repository context via Sourcegraph's search index | No self-serve plan — Free/Pro/Enterprise Starter were discontinued July 23, 2025 |
| Deep Search and Batch Changes address real enterprise-scale problems | Minimum contract starts around $16,000/year, sales-only |
| Supports Claude and OpenAI models, plus bring-your-own-model for compliance needs | No published per-seat price, making budgeting hard before a sales call |
| Multi-IDE support (VS Code, JetBrains, Visual Studio, web) sharing one backend | Individual developers are explicitly redirected to a different product (Amp) |
| Single-tenant deployment and enterprise security controls available | Slower iteration cadence on the consumer-facing editor experience compared with Cursor or Copilot |
Integrations and ecosystem
Cody's ecosystem is built around Sourcegraph's existing enterprise tooling rather than the plugin marketplaces consumer tools lean on. It offers a CLI and full API access for scripting and CI integration, an MCP server for connecting external tools and agents, and ties into Batch Changes, Insights, and Monitoring. There's no Zapier or Slack-app-style consumer integration layer publicized — the integration story here is "plug into your existing Sourcegraph Enterprise deployment," not "connect to your personal productivity stack."
Where it's a strong fit
Cody makes the most sense for organizations that are already Sourcegraph Enterprise customers, or that need an AI assistant which understands relationships across a large, multi-repository codebase rather than a single project. The Deep Search and Batch Changes capabilities are legitimately hard to replicate with editor-native tools like Copilot or Cursor, which mostly reason about the files a developer has open or has recently touched.
Where to think twice
Skip Cody if you're an individual developer, freelancer, or small team — there is no way to buy it without a sales conversation and a five-figure annual commitment, and Sourcegraph itself is pointing that segment of users toward Amp instead. Skip it if you need a transparent, published per-seat price to compare against Copilot's or Cursor's listed plans — Cody Enterprise pricing is quote-only. And skip it if you just want fast, low-friction autocomplete in a single repo; that's a job Copilot, Cursor, or Supermaven-style tools solve with a lighter footprint and a real free or low-cost tier.
Bottom line
Cody's underlying idea — ground an AI assistant in real, indexed knowledge of an entire codebase rather than just open files — is still a technically credible approach, and Deep Search plus AI-assisted Batch Changes are genuinely enterprise-grade capabilities. But Sourcegraph restructured the product in mid-2025 so it's no longer accessible outside an Enterprise contract, removing it from consideration for most developers evaluating AI coding tools on their own budget. If you're already inside a Sourcegraph Enterprise deployment, Cody is worth activating. If not, Copilot, Cursor, or Tabnine will get you a usable AI coding assistant today without a sales call.
FAQ
Does Cody AI have a free plan?
No. Sourcegraph discontinued the Free, Pro, and Enterprise Starter plans on July 23, 2025. Cody is now available only as part of a paid Sourcegraph Enterprise contract.
How much does Cody cost?
Sourcegraph doesn't publish a per-seat price. Enterprise plans start at a minimum annual contract of roughly $16,000, which includes a pool of AI-feature credits that scales with team size, and require contacting sales for a quote.
What happened to Cody's individual Pro plan?
It was sunset alongside the Free plan in July 2025. Existing Pro users were offered a small amount of migration credit toward Amp, Sourcegraph's newer AI coding product aimed at individual developers.
What AI models does Cody use?
Sourcegraph's documentation names Anthropic's Claude models as the default for chat and edits, with OpenAI models available as an alternative, and a DeepSeek-Coder-V2 model for autocomplete. Enterprise customers can also configure their own API keys or self-hosted models.
Which IDEs does Cody support?
VS Code, JetBrains IDEs, Visual Studio, and a web app, all sharing the same Sourcegraph context engine, plus a CLI for scripted use.
Is Cody good for solo developers or small teams?
Not anymore. Without a self-serve plan, Cody isn't practically accessible to individuals or small teams; Sourcegraph directs that audience to Amp instead.
How is Cody different from GitHub Copilot or Cursor?
Cody's core differentiator is pulling context from Sourcegraph's code-search index across an organization's entire set of repositories, plus enterprise-scale tooling like AI-assisted Batch Changes and Deep Search. Copilot and Cursor are built around single-repo, open-editor context and have accessible self-serve pricing that Cody no longer offers.
Can I use Cody if I have data-residency or compliance requirements?
Enterprise plans support self-hosted/bring-your-own-model routing and single-tenant deployment, which can help meet data-residency needs — confirm specifics with sales, since exact compliance certifications aren't detailed publicly.
For broader AI coding options with accessible pricing, see our GitHub Copilot review, Cursor AI review, and Supermaven review, or browse AI & software deals.
