Pricing

Free is not a teaser — it is the private, local version of the same system. Paid plans keep it working when the codebase, team, and memory outgrow one machine.

Free

Local-optimal

$0

Forever. Your code stays local.

Private local code understanding that actually works. No account needed.

  • Local code graph + symbol index
  • MCP context for any coding agent
  • Symbol, literal, and intent-routed search
  • Hot / warm / cold residency
  • Dead-code and reuse reports
  • Local activity memory — stays on your machine

Pro

Scale-optimal

Popular
$19/mo.

Billed annually · $24 monthly

Keeps understanding when your codebase outgrows your laptop.

  • Everything in Free
  • Multi-repo workspaces, cross-repo search
  • Cross-session workspace memory
  • Deeper multi-hop graph context
  • Bounded cloud rerank — explicit, receipt-backed
  • Replay receipts for every improvement
  • Higher indexing limits

Team

Shared memory

$29/user/mo.

Billed annually · $39 monthly

One workspace memory across every member and repo.

  • Everything in Pro
  • Shared workspace memory across members
  • Shared checkpoints and decisions
  • Co-change review signals
  • Governance controls
  • Managed indexing

Enterprise

Governed deployments

Custom

Starting around $18K/year

Govern what agents know and touch — inside your own infrastructure.

  • Everything in Team
  • Private deployment — Docker / VPC
  • SSO, access, and policy controls
  • Audit and compliance
  • Org-level memory governance
  • Priority support

No file-count paywalls and no metered credits. Cloud assist is bounded, explicit, and receipt-backed — the product reports when it was used and why. Cancel anytime.

Questions teams ask before buying

Does my code leave my machine?

Not on Free. Local mode keeps source, the graph index, reports, and memory on your machine — no account required. Paid plans only send what your configured source-upload mode allows: never, metadata_only, or selected_slices. “never” is fully supported; features degrade to local behavior instead of uploading.

Do you train models on our code?

No — Rhei does not train models on customer code by default. Any non-zero-retention provider path must be labeled before it can be enabled for a workspace. See the Data Use overview for the full contract.

What does “bounded cloud assist” mean on Pro?

Cloud rerank or burst is used only for low-confidence cases, is bounded, and every use is reported with a receipt showing when it ran and why. There is no metered credit meter running in the background.

Can we run Rhei entirely inside our infrastructure?

Yes — Enterprise supports private Docker / VPC deployment with SSO, access and policy controls, and audit support, starting around $18K/year. Email [email protected] for a security review and procurement.

How does Team memory respect boundaries?

Workspace memory is scoped per workspace and per member. Knowledge is promoted to the shared layer only by corroboration, decays when contradicted, and stays inspectable with receipts.

How does billing work?

Pro and Team are self-serve with monthly or annual billing — cancel anytime in the customer portal. Team is per user per month. Enterprise is invoiced annually with custom terms.

Security review, DPA, or deployment questions: [email protected] · Full details in the Security overview and Data Use contract.