Comparisons · measured July 2026
“How much is Sourcegraph?” The answer is a stack, not a number.
Ask what enterprise code search costs and you rarely get a price: you get “contact sales.” That is not evasion; it is arithmetic. The bill is built from four things stacked on top of each other (a per-seat license, the server to host it, the compute to keep re-indexing your code, and the people to run it) and each one grows with your team and your repos. This is an honest look at what that price buys, and where a flat-priced local index is the cheaper shape instead.
The short version
Priced per seat, billed on a server.
Enterprise code search is a server-hosted index of an organization's repositories, licensed per developer seat and billed on top of the infrastructure needed to host and continuously re-index them. Sourcegraph is the best-known example: point it at every repo your team works in and it answers cross-repo questions from one place. The reason it costs what it does is the reason it is good: it is built for a whole company, not one laptop.
Sourcegraph is quote-based because the price is a stack (seats × license + server + re-index compute + ops), not a single figure.
What it buys is real: a shared, permission-aware, cross-repo index for a whole team. That is the right tool for org-wide code intelligence.
Every layer compounds (with headcount, with repo size, with churn), which is why the total climbs and never quite sits still.
A local index is a different shape: one flat per-developer price, no server, fresh ~1 ms after a save, feeding 19 AI clients.
They are not rivals. A server answers “what does the company know?”; a local index answers “what is on this machine, for my agent, right now?”
The cost stack, drawn
Four layers you pay for. Only one is the license.
The license is the part you see on the invoice. Underneath it sit three more layers: the server the index lives on, the compute that keeps re-indexing your code as it changes, and the people who deploy, secure, and upgrade it. A local index collapses that whole tower into a single flat per-developer line, because there is no server beneath it to feed.
A recurring fee for every developer who can query the index. Doubles with headcount.
A deployment kept running and warm: their cloud, or servers you stand up yourself.
The server re-crawls and rebuilds as code changes. Big monorepos need real memory and CPU.
Someone deploys, secures, upgrades and monitors it. Quiet, ongoing, and rarely on the quote.
Not a teardown: what the price delivers
What that price genuinely buys.
This is not a hit piece. For a team, a server-hosted code index is the correct tool, and the cost is the cost of doing that job well. When a company's knowledge is shared, its index should be too: one place everyone queries, kept consistent and permission-aware. Here is the value the four layers pay for:
Every repo your team touches, crawled into a single searchable place: one source of truth instead of a dozen local checkouts.
Go-to-definition and find-references that span the whole organization, not just the repo you happen to have open.
Everyone sees the same answers, gated by who is allowed to see what, the thing a local index on your laptop can never be.
Search-and-replace and large-scale batch changes across many repositories at once, a real strength of the server model.
New hires and reviewers explore unfamiliar code from one console, without cloning everything first.
Admin controls, access logs, and governance that a whole company can stand behind.
One caveat baked into all of it: a server sees what has been pushed or synced, not the working copy on your disk this second, and never the files that were never in git. That gap is where a local index lives. More on the two architectures in where the index should live and local vs cloud code search.
Layer by layer
The same cost, from both sides.
This is not “who is cheaper” in the abstract: a server index and a local index buy different scopes. It is a map of what each layer costs you and where a local index simply has nothing to charge for, because there is no server underneath it.
| The cost layer | Cloud code search · Sourcegraph | Local index · Interlinked |
|---|---|---|
| Per-seat license | a recurring fee for every developer who can query | none: the local tool is free |
| Server / hosting | a deployment kept warm: their cloud or one you host | none: it runs on the machine |
| Re-index compute | the server re-crawls as code changes; big repos need memory | rides the live disk, fresh ~1 ms after a save |
| Ops / admin | someone deploys, secures, upgrades and monitors it | one signed installer, no server to run |
| Scope it can see | the pushed / synced repositories | the whole machine, every repo + the non-git 90% |
| What you pay, total | seats × license + servers + compute + ops | free forever; hosting is a separate product from $5.99/mo |
Local-index figures measured on Ryzen 9 9950X3D · 64 GB DDR5 · NVMe · Windows 11 · 4,470,000 files. Sourcegraph enterprise pricing is quote-based (contact-sales), so no single dollar figure exists to quote: the point of the row is the shape of the bill, not a headline number.
The other shape: flat, local, no server
One flat layer. No stack to hold it up.
Interlinked keeps an always-fresh index of every file on the machine (names and contents, across every repo and the non-git 90%) and hands it to whatever agent you use through one connector. There is no server to host, no re-indexing bill on a cluster, no ops rota, and no model anywhere in the box: every result is an index lookup, not a guess. Here is what that costs and delivers, measured.
A single filename lands in 85 µs; across all 4.47M files the median is 139 µs, where Windows Search takes 67 seconds, long enough to refill your coffee, about 480,000× slower. And on install the index seeds a fact no server can represent: replaying 8,455 git events across 12/12 repos in 1.7 seconds, it drew 26,958 relationship pairs and flagged two repos that quietly share 435 identical files. Methodology: Ryzen 9 9950X3D · 64 GB DDR5 · NVMe · Windows 11 · 4.47M files. Full write-ups: the ripgrep comparison and why finding files is the bottleneck.
SOURCEGRAPH FOR THE ORG'S SERVER. INTERLINKED FOR THE MACHINE IN FRONT OF YOU: LIVE, PRIVATE, FLAT-PRICED.
Questions people actually ask
Sourcegraph pricing, answered plainly.
Add the machine in front of you
Keep Sourcegraph for the org. Give your agents the local machine.
One signed installer indexes every file on your machine (names and contents, every repo plus the non-git 90%) and wires up the 19 AI clients you already use. No cloud account, no keys, no JSON editing, and nothing ever leaves the disk. Your team's server index keeps doing its job; every agent on your machine gets a fast, live, private search alongside it, for a flat price with no stack underneath.
Everything local is free, forever: the app, whole-machine search, MCP. Hosting starts at $5.99/mo. Download for Windows · Where should the index live? · Why finding files is the bottleneck