AI Model Picker

AI Model Picker is an open, community-maintained website for choosing an AI model by practical constraints rather than hype. It helps people build a shortlist for coding, writing, image generation, research, chat, and automation.

Status: early MVP. The product and data model are ready for community iteration. The sample catalog is intentionally small, and entries marked needs_review must be checked and updated as providers change their models, pricing, licenses, and policies.

Why this exists

AI model choice is rarely about finding one universal winner. A useful choice depends on the task, budget, hardware, language, API requirements, and privacy constraints. This project makes those trade-offs easier to inspect and discuss.

The project does not use affiliate rankings, fake stars, invented usage counts, or paid placement.

Who it helps

• Developers choosing between hosted APIs and local models
• Teams assessing privacy and deployment constraints
• Creators comparing writing, research, coding, and image tools
• Russian-speaking users looking for an explicit language-support signal
• Open-source contributors maintaining transparent model data

Features

• Searchable AI model catalog
• Filters for task, price, API, open source, local usage, Russian support, and privacy
• Side-by-side model comparison
• Local JSON dataset validated with Zod
• Visible verification dates, source links, and review status
• Responsive interface built with Next.js, TypeScript, and Tailwind CSS
• GitHub issue templates, pull request checklist, tests, and CI

Tech stack

• Next.js App Router
• TypeScript
• Tailwind CSS
• Zod
• Vitest
• GitHub Actions

Run locally

Requirements: Node.js 20.9 or newer and npm.

npm install
npm run dev

Open http://localhost:3000.

Useful checks:

npm run validate:data
npm run lint
npm run typecheck
npm test
npm run build

Add a new model

1. Search data/models.json to make sure the model is not already listed.
2. Add one entry that follows the existing structure.
3. Include at least one primary source: official documentation, model card, repository, or pricing page.
4. Use needs_review when a claim is uncertain or changes frequently.
5. Set last_verified to the date on which you actually checked the sources.
6. Run npm run validate:data and the other checks above.
7. Open a focused pull request using the provided template.

See CONTRIBUTING.md for full data-quality rules.

Data principles

• Claims should be useful, specific, and sourceable.
• Hosted-service privacy and local-deployment privacy are different claims.
• "Open source" can be legally nuanced; document license caveats in notes.
• Unknown is better than an unsupported yes or no.
• Model capabilities, prices, availability, and terms change quickly.

The website exposes the validated catalog at /models.json.

Roadmap

• Improve and verify the starter dataset
• Add per-model detail pages and richer source metadata
• Add saved comparison URLs and shareable filter presets
• Add community-defined evaluation guides
• Add localization, starting with Russian
• Explore automated stale-data reminders without replacing human review

See ROADMAP.md for scope and contribution ideas.

Deployment

The app has no paid API dependency and can be deployed as a standard Next.js project on Vercel, Netlify, or another Node-compatible host. Connect the repository and use the default build command:

npm run build

Community

Please read CONTRIBUTING.md, CODE_OF_CONDUCT.md, and SECURITY.md before contributing.

License

MIT