> ## Documentation Index
> Fetch the complete documentation index at: https://docs.suchi.page/llms.txt
> Use this file to discover all available pages before exploring further.
# Comparison
> Honest side-by-side vs Paperless-ngx, Papra, and docspell — features suchi has, features they have, what's on the roadmap, what's out of scope.
The self-hosted document-management space has real, mature options.
This page compares suchi against the three most credible alternatives
and treats them generously. If your use case fits them better, they
deserve the win — the goal is that a "which one should I pick?"
thread ends with clarity, not a marketing pitch.
**Snapshot date:** 2026-08-05. The alternatives ship faster than
this doc; if you see stale claims about them, please file an issue.
This page is written by the suchi maintainer. It is not third-party
review. We have tried to be honest about the gaps, but read the
alternatives' own docs before deciding.
## The four projects, one line each
| Project | One-line | License |
| ----------------- | ------------------------------------------------------------------------------- | -------- |
| **suchi** | Go single-binary DMS with SQLite-first storage + plugin seams | AGPL-3.0 |
| **Paperless-ngx** | The reference OSS DMS. Django + Redis + Postgres + Celery. Ten years of polish. | GPL-3.0 |
| **Papra** | Minimal, opinionated document management. Node/TypeScript, fast to try. | AGPL-3.0 |
| **docspell** | Scala + Postgres + Solr. Powerful indexing, joex worker split. | GPL-3.0 |
## The frame
* **suchi** is fighting for "boring underneath, easy to get around."
One Go binary, one SQLite file, no external services. If you value
"I can grep the whole codebase" and "backup is `tar czf`", this is
the pitch.
* **Paperless-ngx** is fighting for feature completeness and
ecosystem gravity — mobile apps, integrations, community, guides.
If you want the option that comes up first on Reddit and has an
answer to almost any question already documented, this is it.
* **Papra** is fighting for "modern UX, minimal to try, doesn't
demand you learn a stack." Slick UI, quick onboarding.
* **docspell** is fighting for indexing depth and language coverage.
If you have a large multilingual corpus and need serious search,
this is it.
## Feature matrix
Legend: ✅ = ships · 🚧 = roadmap · ➖ = not planned · ❓ = unknown
### Ingest
| Feature | suchi | Paperless-ngx | Papra | docspell |
| ------------------------------ | :---------------------------------------------------: | :-----------: | :---: | :-----------: |
| Upload API | ✅ | ✅ | ✅ | ✅ |
| Watched folder | ✅ | ✅ | ✅ | ✅ |
| Email intake (IMAP) | 🚧 (scaffold + polling loop; needs real mailbox test) | ✅ | ➖ | ✅ |
| Sidecar-JSON ingest spec | ✅ | ➖ | ➖ | ➖ |
| Multi-doc split on scan | ✅ | ✅ | ➖ | ➖ |
| Blank-page removal | ✅ | ✅ | ➖ | ➖ |
| Pre-consume script hook | ✅ | ✅ | ➖ | ➖ |
| OCR (scanned PDFs, images) | ✅ (tesseract + ocrmypdf) | ✅ | ➖ | ✅ (5 engines) |
| Office documents (docx/xlsx/…) | ✅ (via anydoc) | ✅ | ➖ | ✅ |
| Email (.eml) → child docs | ✅ | ➖ | ➖ | ➖ |
### Metadata + classification
| Feature | suchi | Paperless-ngx | Papra | docspell |
| ---------------------------------- | :-----------------------: | :-----------: | :---: | :------: |
| Tags | ✅ (nested via parent\_id) | ✅ | ✅ | ✅ |
| Correspondents (multi per doc) | ✅ | ➖ (one) | ➖ | ✅ |
| Document types | ✅ | ✅ | ➖ | ✅ |
| Storage paths (Gonja templates) | ✅ | ✅ | ➖ | ➖ |
| Custom fields (typed) | ✅ | ✅ | ➖ | ✅ |
| Rules-engine classifier | ✅ | ✅ | ➖ | ➖ |
| LLM classifier (OpenAI-compatible) | ✅ | ➖ | ➖ | ➖ |
| Johnny.Decimal tree by default | ✅ | ➖ | ➖ | ➖ |
| Trigger→conditions→actions rules | ✅ | ✅ | ➖ | ➖ |
| Human-in-the-loop approvals | ✅ | ➖ | ➖ | ➖ |
### Surface
| Feature | suchi | Paperless-ngx | Papra | docspell |
| ----------------------------------------------------------- | :------------------: | :-------------------: | :-------: | :------: |
| REST API | ✅ | ✅ | ✅ | ✅ |
| OpenAPI schema at `/api/schema/` | ✅ | ➖ | ❓ | ✅ |
| DRF pagination shape | ✅ | ✅ | ➖ | ➖ |
| Mobile-app wire compat (paperless-mobile / swift-paperless) | 🚧 | ✅ | ➖ | ➖ |
| MCP server (agent tools) | ✅ (`suchi mcp`) | ➖ | ➖ | ➖ |
| Share links (expiry + password) | ✅ | ✅ | ✅ | ➖ |
| Full-text search | ✅ (SQLite FTS5) | ✅ (Postgres tsvector) | ✅ | ✅ (Solr) |
| Search filters (tags/corr/type/…) | ✅ | ✅ | ✅ | ✅ |
| Autocomplete | ✅ | ✅ | ➖ | ✅ |
| Web UI | ✅ (Oat + minimal JS) | ✅ (Angular) | ✅ (React) | ✅ (Elm) |
### Ops
| Feature | suchi | Paperless-ngx | Papra | docspell |
| --------------------------------- | :--------------------------: | :----------------------: | :---------: | :-------------------: |
| Single binary | ✅ | ➖ | ➖ | ➖ |
| SQLite as primary store | ✅ | ➖ (Postgres) | ✅ | ➖ (Postgres) |
| External services required | none | Postgres + Redis | none | Postgres + Solr |
| Runtime deps at boot | libc + optional binaries | Python + workers + Redis | Node | JVM + Postgres + Solr |
| Signed images + SBOM | ✅ | ➖ | ➖ | ➖ |
| Import from Paperless-ngx | ✅ (`suchi import paperless`) | (self) | ➖ | ➖ |
| Audit log | ✅ | ➖ | ➖ | ➖ |
| Multi-user + per-user ownership | ✅ | ✅ | ✅ | ✅ |
| ACLs (groups + object perms) | 🚧 (Phase 6) | ✅ | ➖ | ✅ |
| Idle RAM (measured, small corpus) | 🚧 (benchmark pending) | \~350-500MB | \~100-150MB | \~500MB-1GB |
## Where each project genuinely wins
### Paperless-ngx wins if…
* You want the biggest community and the most Reddit answers.
* You need the fully mature mobile apps working out of the box today
(suchi's mobile compat is 90% shipped but not yet green against
pinned mobile release tags — that's the last Phase 4 batch).
* You want an option so battle-tested that "will it work?" is a
non-question. Ten years of production is worth a lot.
* Your infrastructure already includes Postgres + Redis and adding
another service is fine.
### Papra wins if…
* You care about "prettiest out of the box" and "smallest cognitive
load to try."
* Your requirements are metadata-light: tags + folders + share
links are enough.
* You want a smaller feature surface — less to configure, less that
can break.
* You value the Node/TypeScript ecosystem for extension.
### docspell wins if…
* You have a **multilingual** corpus and need real search across
languages that FTS5 doesn't tokenize well.
* You want the joex worker-split model at scale.
* You already run a JVM stack.
* You need Solr's advanced indexing (proximity, boosting, field
weighting).
### suchi wins if…
* You want one Go binary, one SQLite file, no external services.
* You self-host on a NAS / Pi / small VPS where "keep it small"
matters.
* You value the maintainer surface: grep the whole codebase, one
language, no ORMs.
* You want the Johnny.Decimal taxonomy shipping opinion out of the
box, or the option to run flat if you don't.
* You want MCP-native agent tools without a translation layer.
* You need multi-party correspondents per doc (invoice sender vs.
recipient) as first-class data.
* You want an audit log for "who deleted that?" from day one.
* You want mobile-app wire compat as a design goal so the existing
apps work against a smaller substrate.
* You value approvals + automations as distinct, well-named
primitives — not one confused feature.
## Where suchi doesn't win
Being honest about the gaps:
* **Community.** Paperless-ngx has years of accumulated Reddit,
guides, videos. Suchi is pre-1.0.
* **Mobile compat green.** The endpoints exist, but the
golden-transcript replay against pinned app releases isn't run
yet. Paperless-ngx is what the apps target natively.
* **UI depth.** Paperless-ngx's Angular UI has features suchi
doesn't (bulk edit, saved-view editor UI, drag-drop custom-field
editor). Suchi's UI is intentionally thinner.
* **Ecosystem plugins.** Paperless has more community-maintained
side-projects (mobile scan apps, integrations).
* **Indexing depth.** SQLite FTS5 is great for most Latin-alphabet
corpora; not the equal of Solr for multilingual heavyweight use.
## Roadmap items worth flagging
* **Idle-resource benchmark** — planned as a published measurement,
not a claim. Until it's run, the RAM/CPU numbers in the "Ops"
table above are approximate from public issue-tracker reports.
* **Public demo instance** — pending the repo flip.
* **Postgres backend plugin** — E4 in the plan, triggered by
measured need (sustained p95 write > 500ms).
## "But can I move if I chose wrong?"
* **Paperless-ngx → suchi**: `suchi import paperless ` is a
first-class product feature. See [importer](/importer).
* **suchi → Paperless-ngx**: no shipped export tool. The DB is
SQLite; the schema is documented. You'd write a script.
* **suchi → docspell / Papra**: same as above.
If unsure between suchi and Paperless-ngx and you already have data
in Paperless-ngx, the import path being one-way in the other
direction is worth weighing. If you're greenfield, this is not a
concern.
## Not compared here
* Commercial DMS products (M-Files, DocuWare, etc.). Different
market — enterprise procurement + support contracts.
* SaaS document platforms (Papra Cloud, Google Drive, Notion).
Different tradeoff — you get zero-ops in exchange for the platform
reading everything you upload.
* Immich / Photoprism. Adjacent but photo-first, not document-first.
## Update history
This page is manually maintained. If any claim is wrong, file an
issue with a link to the alternative's docs and we'll fix it.
* **2026-08-05** — initial version