A solo developer posted to Hacker News on July 20, 2026 announcing a free AI platform called 'Use AI Free' that supports an impressive range of interfaces and integrations. The project offers GUI, web, CLI, and Telegram access from the get-go—a multi-platform approach that's rare for side projects from individual contributors. With 2,000 MCP (Model Context Protocol) servers supported alongside 100,000 skills, this is a substantial toolkit that could appeal to developers tired of juggling separate tools for different AI workflows.

Under the Hood: LiteLLM and Ollama Integration

The platform leans on LiteLLM as its backbone, which enables compatibility with over 100 AI providers through a unified interface. This means users aren't locked into a single vendor—they can swap between models from OpenAI, Anthropic, Google, and dozens of others without changing their code or workflow. For privacy-conscious developers, the project also supports running local models via Ollama, keeping sensitive data entirely within the user's infrastructure rather than bouncing requests to third-party APIs.

Localization and Accessibility

The /lang command works across 34 languages, making this platform accessible to non-English speakers who often get underserved by developer tooling. The 'Mesa Redonda' feature—Spanish for 'round table'—appears to be a collaboration or discussion mode built into the system. Voice support and OCR capabilities round out the feature set, giving users multiple ways to interact with AI models beyond traditional text prompts.

Embedded Sandbox OS

One standout inclusion is an embedded sandbox OS environment where developers can safely test prompts, skills, and integrations without risking their main system or API quotas. This kind of safety net matters when you're experimenting with new LLM configurations or testing third-party MCP servers that might behave unpredictably.

Key Takeaways

  • Supports GUI, web, CLI, and Telegram interfaces out of the box
  • LiteLLM integration connects to 100+ AI providers; Ollama enables fully local inference
  • 2,000 supported MCP servers and 100,000 skills offer massive extensibility
  • Built-in sandbox OS, voice support, OCR, and multilingual /lang command in 34 languages

The Bottom Line

This is a ambitious solo project that tries to be everything-to-everyone for AI access. Whether it can maintain quality across 100K skills remains to be seen, but the architecture choices—LiteLLM abstraction plus local Ollama support—show solid engineering instincts. Worth watching if you want a unified interface that doesn't force vendor lock-in.