paperless-gpt vs Unstract
Short answer
paperless-gpt: LLM add-on for paperless-ngx that does better OCR and writes titles, tags and correspondents. Unstract: No-code LLM platform that turns invoices, forms and other documents into structured JSON. Both are in our Document processing category.
Choose paperless-gpt if you want: fewer license conditions (Permissive vs Network copyleft (AGPL)).
Choose Unstract if you want: larger community (7.3k vs 2.8k GitHub stars).
paperless-gpt publishes no official RAM figure, so we cannot compare hosting cost directly.
Last updated · computed from each project's official docs and live GitHub data · sources
paperless-gpt vs Unstract side by side
| paperless-gpt | Unstract | |
|---|---|---|
| Category | Document processing | Document processing |
| What it is | LLM add-on for paperless-ngx that does better OCR and writes titles, tags and correspondents. | No-code LLM platform that turns invoices, forms and other documents into structured JSON. |
| License | MIT | AGPL-3.0 |
| License type | Permissive | Network copyleft (AGPL) |
| License conditions | Permissive. Internal and commercial use allowed; keep the license notice. | Free for internal and commercial use. If you modify Unstract and let users interact with it over a network, you must offer them the modified source code under AGPL-3.0. |
| Status | Active | Active |
| GitHub stars | 2.8k | 7.3k |
| Last commit | Oct 16, 2026 | Oct 26, 2026 |
| Latest release | v0.31.0 | v0.198.0 |
| Minimum hardware | Not published | 8 GB RAM |
| Recommended hardware | Not published | Not published |
| RAM measured by us | Not tested yet | Not tested yet |
| GPU | Not needed for the app itself. A GPU only matters if you run the LLM locally (for example with Ollama). | Not required by the platform itself; LLMs are reached through the provider you configure (cloud API or a local server such as Ollama). |
| Est. server cost | Not estimated | from €35.99/mo |
| Deploy method | Docker Compose | Docker Compose (run-platform.sh) |
| Known gotchas | 3 listed | 4 listed |
| Telemetry | Not documented | On by default |
| Calls home by default | Not documented | Not documented |
| SSO (OIDC / SAML / LDAP) | Not supported (OIDC not supported · SAML not supported · LDAP not supported) | Paid / enterprise only (OIDC paid · SAML paid) |
| Audit log | Not documented | Not documented |
| RBAC | No | Not documented |
| Details | Full paperless-gpt guide → | Full Unstract guide → |
Key differences
- License: paperless-gpt: MIT (Permissive). Unstract: AGPL-3.0 (Network copyleft (AGPL)).
- Official hardware: paperless-gpt: not published. Unstract: minimum 8 GB RAM.
- Est. server cost: paperless-gpt: not estimated (no official RAM figure). Unstract: from €35.99/mo.
- Deploy: paperless-gpt: Docker Compose · multi-container stack. Unstract: Docker Compose (run-platform.sh) · multi-container stack.
- GitHub stars: paperless-gpt: 2.8k. Unstract: 7.3k.
- Last commit: paperless-gpt: Oct 16, 2026. Unstract: Oct 26, 2026.
- Latest release: paperless-gpt: v0.31.0 (Oct 14, 2026). Unstract: v0.198.0 (Oct 26, 2026).
Same for both: Category: Document processing; GPU: optional / see notes; Status: Active.
paperless-gpt vs Unstract: FAQ
paperless-gpt or Unstract: which should a small business choose?
Choose paperless-gpt if you want: fewer license conditions (Permissive vs Network copyleft (AGPL)). Choose Unstract if you want: larger community (7.3k vs 2.8k GitHub stars). paperless-gpt publishes no official RAM figure, so we cannot compare hosting cost directly.
Which is cheaper to self-host, paperless-gpt or Unstract?
paperless-gpt: not estimated (no official RAM figure). Unstract: from €35.99/mo. Our estimate is the cheapest Hetzner Cloud plan that meets each project's official RAM and vCPU figure; it covers the server only.
How much hardware do paperless-gpt and Unstract need?
paperless-gpt: not published. Unstract: minimum 8 GB RAM.
What is the license difference between paperless-gpt and Unstract?
paperless-gpt: MIT (Permissive). Permissive. Internal and commercial use allowed; keep the license notice. Unstract: AGPL-3.0 (Network copyleft (AGPL)). Free for internal and commercial use. If you modify Unstract and let users interact with it over a network, you must offer them the modified source code under AGPL-3.0. This is a summary, not legal advice.
Is paperless-gpt or Unstract more actively maintained?
paperless-gpt: Active, last commit Oct 16, 2026. Unstract: Active, last commit Oct 26, 2026. GitHub data refreshed Oct 27, 2026.
Which is easier to deploy, paperless-gpt or Unstract?
paperless-gpt installs via Docker Compose (multi-container stack); Unstract via Docker Compose (run-platform.sh) (multi-container stack). paperless-gpt lists 3 known gotchas and Unstract lists 4; see each app page.
Related
- paperless-gpt self-hosting guide · Unstract self-hosting guide
- All paperless-gpt alternatives, ranked
- All Unstract alternatives, ranked
- MinerU vs paperless-gpt
- MinerU vs Unstract
- PaddleOCR vs paperless-gpt
- PaddleOCR vs Unstract
- Paperless-AI vs paperless-gpt
- Paperless-AI vs Unstract
- Add a third app in the interactive comparison
Sources
- paperless-gpt GitHub repository: github.com/icereed/paperless-gpt
- paperless-gpt documentation: github.com/icereed/paperless-gpt#getting-started
- paperless-gpt license: github.com/icereed/paperless-gpt/blob/main/LICENSE
- paperless-gpt hardware requirements: github.com/icereed/paperless-gpt#prerequisites
- paperless-gpt deploy instructions: github.com/icereed/paperless-gpt#docker-compose
- Unstract GitHub repository: github.com/Zipstack/unstract
- Unstract documentation: docs.unstract.com/unstract/unstract_platform/quick_start
- Unstract license: github.com/Zipstack/unstract/blob/main/LICENSE
- Unstract hardware requirements, Unstract deploy instructions: github.com/Zipstack/unstract#readme
- Hetzner Cloud pricing used for cost estimates (read 2026-10-25): www.hetzner.com/cloud/regular-performance
- GitHub REST API (stars, commits, releases): docs.github.com/en/rest
Every value is generated from the same verified data as the app pages. Spotted something wrong? Tell us.