Ollama — system requirements

Runs and serves large language models locally via a CLI, API, and Docker image; footprint scales with the loaded model. · website · repository

No official figure published for: minimum RAM, recommended RAM, minimum CPU cores, recommended CPU cores

Ollama publishes no base/runtime RAM or CPU figure for the app itself — actual memory use is set entirely by which model is loaded (per-model sizes are listed on individual ollama.com/library model pages, not as a single app-level spec). Upstream issue #2418 ('What are the system requirements?', closed without a docs resolution — no linked PR or requirements page) requested this documentation for disk/RAM/GPU; the docs/ tree (faq.mdx, gpu.mdx, linux.mdx, etc., checked 2026-08-23) still has no general RAM/CPU minimum. Per-model tiers are out of scope for this entry — encoding one model's footprint as the app's min/rec figure would misrepresent scope (Open WebUI precedent, Defect Class #2/#3).

We publish nothing we can't source. Evidence of the gap.

External services

No required external services (source)

Container

Image: ollama/ollama · compressed size (amd64): 3223 MB · architectures: amd64, arm64 · source, retrieved 2026-08-23

Related apps

ONLYOFFICE Document Server — Self-hosted online office suite for collaborative editing of documents, spreadsheets, and presentations.
Open WebUI — Self-hosted web interface for interacting with local and remote large language models.
OpenProject — Self-hosted project management suite with Gantt charts, agile boards, and time tracking.