Change language
Get Started

Pricing

Distributed individually.

every deployment ships the full binary — 49 tools, 9 crates, no usage metering

StarterFor the individual researcher who wants the whole toolbox on one machine.
Custom per deployment
Seats 1Tools All 49, no meteringSearch 7 backends, smart fusionMemory Local SQLite + embeddingsSupport Community
Request access
TeamShared contact database, CRM sync and API access for a small research team.
Custom per deployment
Seats Up to 10Contacts PostgreSQL ContactDbCRM amoCRM · Bitrix24 · HubSpotAccess HTTP API + dashboardSupport PrioritySLA 8×5
Request access
EnterpriseOn-prem isolation, compliance and extensions for regulated workloads.
Custom per deployment
Seats CustomDeploy On-prem · multi-process isolationCompliance ZDR · SSO · DPAExtend Custom tools + MCP serversSupport Dedicated engineerSLA 24×7
Request access

Air-gapped deployment

Run the full pipeline with zero external calls: local models over an OpenAI-compatible endpoint, offline TF-IDF embeddings, and contacts that never leave your perimeter.

Zero Data RetentionSSO / SAMLDPAOn-premCustom models
Talk to us →

Compare

Same binary, different scale

Every tier ships all 49 tools — plans differ in seats, storage, integration depth and support, never in capability.

FAQ

Questions, answered

Parallel Research is a closed product distributed individually — no public package registries or app stores. You receive a single static binary containing the CLI, TUI, HTTP server, dashboard, MCP server and worker processes. Every deployment is complete and identical: 49 tools, 9 crates, no usage metering.

No. Everything — contacts, memory, jobs, reports — lives on your own infrastructure: local SQLite or your PostgreSQL. Nothing leaves your machines unless you push it to your own CRM, and there is no telemetry on your data.

Any OpenAI-compatible endpoint: seven providers work out of the box, and you can plug in self-hosted or custom models. Per-role model routing lets each agent persona run on its own model.

Yes. Offline TF-IDF embeddings (512 dimensions, no network), local models behind an OpenAI-compatible endpoint and on-device processing let the full pipeline run without any external calls.

Request your deployment

hello@parallel.research →