Catervas tools library

Ollama

Runs AI models privately on your own computer, so your data stays in your office.

AI capabilities

Runs open models (Llama, Gemma, Qwen, DeepSeek, Mistral, etc.) fully on your own hardware. Exposes an OpenAI-compatible API, so other tools can use it instead of a paid cloud AI key.

Pricing

Free license. Hosting estimate: $0 (runs on a PC you own; a 16 GB+ RAM machine or a GPU helps)

Open source

  • Code: https://github.com/ollama/ollama
  • GitHub stars: 186021
  • Latest release: 2026-11-03
  • Hosting: Runs on your own Mac, Windows or Linux PC or office server (one-click installer). No hosting needed. Optional paid Ollama Cloud models.
  • Backing: Company-backed (Ollama Inc.). Extremely active: v0.40.14 on 2026-11-03.
  • The runtime is MIT. Ollama also sells optional cloud model access (Pro $20/mo, Max $100/mo, Team $500/mo). That is not needed to run models locally for free.
  • No GitHub repo advisories, but 2026 CVEs exist. CVE-2026-7482 (critical, GGUF loader heap read) is fixed in 0.17.1. CVE-2026-5757 'Bleeding Llama' (unauthenticated heap leak via the quantization engine) and CVE-2026-42248 (Windows auto-updater skips signature checks, reported up to 0.17.5) have no vendor-confirmed fix version that we could find. Mitigation: run the latest version, keep the API on localhost (the default, port 11434), and never expose it to the internet.

Trust checks

  • Admin-verified listing: Pass · Reviewed by the Catervas team: links and prices checked Nov 5, 2026
  • License verified: Pass · MIT is on the license allow-list
  • Active maintenance: Pass · Last release or commit 2026-11-03
  • No open critical security advisories: Fail · Two 2026 vulnerabilities have no vendor-confirmed fix. Keep it on localhost and never expose it to the internet.
  • Privacy and data policy: Pass · Self-hosted: your data stays on your own hosting.
  • Company or foundation backing: Pass · Company-backed (Ollama Inc.). Extremely active: v0.40.14 on 2026-11-03.

Problems it helps with

  • Data security, WISP rules and phishing risk: Keeps client and patient data in the office. A tech helper installs it once on an office PC and points AnythingLLM (document Q&A) or Phlox (AI scribe) at it. Nothing is sent to OpenAI or another cloud, which simplifies WISP and HIPAA data-handling questions.
  • Charting overload and after-hours notes: Keeps client and patient data in the office. A tech helper installs it once on an office PC and points AnythingLLM (document Q&A) or Phlox (AI scribe) at it. Nothing is sent to OpenAI or another cloud, which simplifies WISP and HIPAA data-handling questions.

Data as JSON: /api/tools/ollama · Compare tools: /compare · llms.txt

Information, not professional advice. Rankings can't be bought.