Outlier vs Sourcegraph Cody: local coding AI vs enterprise cloud AI

Disclosure up front: I'm Matt Kerr, and I build Outlier, so read this as a comparison written by an interested party. Every Cody claim comes from Sourcegraph's own docs, linked at the bottom; every Outlier number is measured. The two barely overlap in shape.

What Cody is in mid-2026

Sourcegraph's docs call Cody an AI coding assistant that uses recent LLMs plus your development context to help you understand, write and fix code faster. The page carries a badge reading "Supported on Sourcegraph Enterprise," and that badge is the whole story.

On June 25, 2025, Sourcegraph announced that Cody Free and Cody Pro would shut down on July 23, 2025 — a year ago today — and that Enterprise Starter would no longer include Cody. Departing users were moved to Amp with $10 and $40 in credits; Enterprise customers, the post said, were unaffected. Both install guides now list one prerequisite: "A Sourcegraph Enterprise account with Cody enabled."

The clients are still maintained — VS Code v1.155.0 shipped 2026-06-17 with 868,377 installs and a 3.98/5 average. But Cody is absent from Sourcegraph's homepage, pricing page and 2026 changelog, and unmentioned in the December 2, 2025 announcement that Sourcegraph and Amp were splitting into separate companies.

Where each one lives

Cody is an extension first. Documented clients are VS Code, JetBrains IDEs, Visual Studio (Experimental), the web app, and a CLI you install with npm install -g @sourcegraph/cody on Node.js v20+ — itself "in the Experimental stage for Enterprise accounts." No standalone desktop or mobile client appears in that list.

Outlier is the opposite shape: a native macOS app for Apple Silicon (M1 or later, macOS 12+). No Intel build, no Windows, no Linux, no mobile app — and it isn't an editor extension. If you want ghost-text completion in your gutter as you type, Cody does that and Outlier doesn't.

Which model does the work

Cody routes to frontier cloud models: Claude Opus 4.8 through 4.5, Sonnet 5/4.6/4.5 and Haiku 4.5; Gemini 2.5 Pro, 3.1 Pro and 3.5 Flash; GPT-5.4 through GPT-5, GPT-4o and o3. Default chat is Sonnet 4.5. Autocomplete uses DeepSeek V2 Lite Base and an Autoedits fine-tune of DeepSeek Coder V2 via Fireworks.ai — open weights, consumed as a hosted API rather than run by you.

Outlier ships seven tiers you download once and run locally, from Nano 4B (2.37 GB) up to Plus 397B-a17b (209 GB, streaming experts from SSD at ~11 GB peak RSS). The coding tiers are Core and Code 27B: 15.13 GB, 24 GB of RAM, 20.7 tok/s on an M1 Ultra, weights published at huggingface.co/Outlier-Ai. Cody's client source isn't public any more: github.com/sourcegraph/cody 404s, and the Apache-2.0 cody-public-snapshot repo is archived, last pushed 2025-08-01.

Does your code leave the machine?

With Cody, always. The FAQ is explicit that inference is remote: Cody sends code snippets of up to 28 KB per request to a third-party cloud LLM, and a self-hosted Sourcegraph instance still needs internet access for those calls, routed by default through Cody Gateway.

The written guarantees are strong. The Cody notice states Sourcegraph and its partner LLMs don't use code from Cody Enterprise or Pro teams to train models, and that partner LLMs retain no input or output, embeddings included, beyond the time it takes to generate the output. Fine-tuning is opt-in, and enterprises can bypass the Gateway with their own infrastructure. One caveat: the only documented local-inference path is the Ollama page, which still says support is "available for Cody Free and Pro plans" — plans switched off a year ago.

Outlier's answer is structural rather than contractual. Models run on your Mac, no account, no usage cap, Wi-Fi off if you like. There's no data policy to read because there's no transfer.

Cost, and the side-by-side

Cody has no self-serve purchase and no published per-seat price — it appears nowhere on sourcegraph.com/pricing, where the only figure is an Enterprise plan "Starting at $16K." Outlier is free for Nano and Lite with no account, $20/month or $149/year for all seven tiers, lifetime from $99.

The coding receipts, honestly

Here's my side, measured rather than lifted from a leaderboard. On a blind slice of SWE-bench Verified, Outlier's local 27B resolved about 45% — 18 of 40. Code 27B scores 0.866 on HumanEval. Core 27B also matched Claude Opus on 98.9% of rubric checks across a 54-prompt head-to-head, and 100% on the nine hardest.

Those are my own runs on my own hardware — not a leaderboard submission, not audited, and 40 instances is a small slice. Forty-five percent is a real result for a 15 GB model running with the network off. It's also below what the strongest cloud coding agents resolve; on a hard multi-file bug, Sonnet 4.5 or Opus 4.8 lands the fix more often than a local 27B does. Sourcegraph doesn't publish a resolve rate for Cody that I could find, so there's no number to set beside mine.

The context gap is starker than the model gap. Cody builds context from keyword search, the Sourcegraph Search API against remote repositories, and the code graph, plus @-mentions of files, symbols and repos — so it can answer questions about code that has never touched your laptop. Outlier can't.

Where Sourcegraph Cody genuinely wins

Cody's moat is real, and a local Mac app cannot cross it.

• Organization-scale context. Retrieval spans every remote repository the company has indexed. For a monorepo or a 5,000-repo org, that's context Outlier can't see, because the code isn't on the machine.
• Permission-aware retrieval. The FAQ states retrieval ensures only code the user has read permission for is retrieved — a hard problem a single-user tool never has to solve, and a hard requirement in regulated environments.
• No hardware ceiling. Opus 4.8 and GPT-5.4 don't care about your RAM, quantization or thermals, and the model list refreshes without re-downloading tens of gigabytes.
• Written data guarantees. Contractual no-training on Enterprise code, zero retention beyond generation, opt-in fine-tuning, plus bring-your-own-key if you'd rather the vendor never see prompts.
• Enterprise governance. RBAC role gating, admin-chosen models, Context Filters for sensitive paths, an org-wide Prompt Library. A single-user desktop app has no answer to that.
• One assistant across a mixed toolchain — VS Code, eleven JetBrains IDEs, Visual Studio and the web — with nothing to provision locally. It behaves the same on an 8 GB laptop as on a workstation; Outlier does not.

Who should pick which

Pick Cody if your company already runs Sourcegraph, your code lives in hundreds of repositories nobody clones locally, you need permission-scoped retrieval and admin governance, and you want frontier-model quality on hard multi-file work. Its differentiator is the Sourcegraph index — without an indexed deployment you're paying enterprise prices for an ordinary IDE chat panel.

Pick Outlier if you're an individual or small team on Macs, your code needs to stay put by architecture rather than by contract, you work offline, or you simply can't buy Cody — which, since Free and Pro were retired, is most people.

And my side's gaps: no web search, no image generation, no mobile app, no voice mode, no Windows or Linux build, no cloud sync. The coding tiers want 24 GB of RAM, and Outlier can't see a repository that isn't on your disk.