Pico AI Server for MLX LLM VLM 17+

Works with Ollama & OpenAI

Starling Protocol Inc

    • Free

Screenshots

Description

Super-fast LLM & VLM server for home and office networks. Powered by MLX with DeepSeek, Gemma, Llama, Mistral, Qwen, and Phi models.

The #1 Ollama and OpenAI-compatible LLM and VLM server for macOS. Run DeepSeek, Llama, Gemma, and more—all on your Mac. No cloud. No rate limits. Full privacy.

• Instant AI Server
Turn your Apple silicon Mac into a lightning-fast Ollama and OpenAI-compatible LLM server for your local network with just one click. No terminal needed.

• Ollama and OpenAI-compatible
Pico supports Ollama-style and OpenAI-compatible APIs, plus all your favourite chat clients like Bolt AI, Open WebUI, Apollo, Msty, and more.

• Private & Offline by Default
Every prompt, every reply, every model stays on your Mac. Works even without internet.

• Built for Apple Silicon Speed
MLX-accelerated to harness the full power of M-series chips—up to 2-3× faster than standard ports.

• 300+ Ready-to-Use Models
Coding assistants, creative writers, research tools—whatever you need. Switch models as easily as changing tracks, or import your own MLX models.

• Friendly Dashboard, Pro Controls
Adjust creativity, context, system prompts and more with simple sliders. Advanced settings for power users.

––– REQUIREMENTS –––
• Apple Silicon Mac (M-series)
• 16 GB RAM minimum
• 32 GB+ recommended for larger models

NO SUBSCRIPTIONS • NO SIGN-UPS • NO CLOUD

Download Pico today and give your home or team a private, high-performance AI server—powered entirely by your Mac.

• Pico AI Homelab supports over 300 leading LLM and VLM models, including:
• Google Gemma 3
• DeepSeek R1
• Meta Llama
• Alibaba Qwen 3
• Alibaba QwQ
• Polaris
• Microsoft Phi 4
• Microsoft BitNet
• Mistral
• DeepHermes
• Granite Code
• Hugging Face SmolLM
• Hugging Face SmolVLM
…And many more

• Pico AI Homelab supports 23 embedding models, such as:
• BERT
• RoBERTa
• XLM-RoBERTa
• CLIP
• Word2Vec
• Model2Vec
• Static

• Compatible with your favourite chat apps, including:
• Open WebUI
• Apollo AI
• Bolt AI
• IntelliBar
• Msty
• Ollamac
• MindMac
• Enchanted
• Kerling
• LibreChat
• Hollama
• Ollama-SwiftUI
• Witsy
• Reactor AI
... And many more

What’s New

Version 1.1.18

- Bug fixes:
- Resolved “400: model not found” error in Open WebUI
- Fixed Open Chat client for Pico on non-default ports
- Embeddings endpoint now returns embeddings as expected
- New models:
- Gemma 3 QAT
- Polaris 4B Preview
- Microsoft BitNet

App Privacy

The developer, Starling Protocol Inc, indicated that the app’s privacy practices may include handling of data as described below. For more information, see the developer’s privacy policy.

Data Not Collected

The developer does not collect any data from this app.

Privacy practices may vary based on, for example, the features you use or your age. Learn More

More By This Developer