Thank you for considering contributing.
This list curates open-source tools and resources for running AI locally — on your own hardware, with your own data, without cloud dependency. Submissions are evaluated against that principle.
A resource is eligible if ALL of the following are true:
- Locally executable. Runs on the user's hardware. Cloud APIs, SaaS platforms, and remote-only services do not qualify, even if they expose an "API" that resembles local tools.
- Open source OR free-as-in-beer. Source available under a recognized open-source license (MIT, Apache 2.0, GPL, BSD, etc.), OR free binary distribution with no paid tier required for core functionality.
- Actively maintained. Commits, releases, or issue activity within the past 12 months.
- Genuine local AI relevance. Tools, libraries, frameworks, models, guides, or communities that directly serve people running AI locally.
- Inference engines and runtimes (llama.cpp, Ollama, vLLM, etc.)
- Model repositories with downloadable weights
- Local AI tools and applications (agents, chat UIs, fine-tuning frameworks)
- Hardware guides and benchmarks
- Setup tutorials and optimization guides
- Communities focused on local AI (Reddit, Discord, forums)
- Quantization tools and formats
- Cloud-only services or APIs. OpenAI, Anthropic API, Google Gemini API, etc. — even if they have "local-like" SDKs.
- Subscription resale, account sharing, or pooled-account services. Services that resell ChatGPT, Claude, or other paid subscriptions to multiple users.
- Paid SaaS without genuine free tier. Free trials, "freemium" with crippled core functionality, and "contact us for pricing" services don't count as free.
- Self-promotional content without technical substance. Blog posts that exist primarily to drive traffic to commercial services.
- Closed-source commercial products without locally-running open-source equivalents.
- Affiliate-link aggregators or coupon sites.
- Abandoned projects (no activity in 12+ months).
- Duplicate entries.
- Resources requiring account registration to access core functionality (gating contradicts the local-AI principle).
- Fork the repository
- Add your resource in the appropriate alphabetized section
- Use the format:
- [Name](URL) - Brief description (under 100 chars) - Verify the link works and the project meets the eligibility criteria
- Submit a pull request
Pull requests are strongly preferred over issues. Issues requesting addition without an accompanying PR will typically be closed without action.
- Link to the primary source (official GitHub repo, official site, or canonical project page — not aggregators or marketing landing pages)
- Description should explain WHY the resource is useful, not WHAT it is
- Active maintenance verified (recent commits, working install, responsive maintainers)
- Entries alphabetized within sections
- Resources fitting multiple categories: place in the most relevant single category
- New sections require 3+ qualifying resources to justify
The following patterns are closed silently:
- Outreach via email or external messaging asking for "collaboration" or "partnership" instead of opening a PR
- Issues that ask for our email address before stating what the resource does
- Generic "would love to collaborate" messages without specifics
- Resources that violate the eligibility criteria above
- Resources that exist primarily to drive traffic elsewhere
Add resources via PRs. Generic outreach gets closed.
For genuine eligibility questions: open an issue clearly stating the resource, the license, and how it meets the locally-executable criterion.
For everything else: just submit a PR. We review on merit, not on prior conversation.