Open Source on Hugging Face

Sovereign AI Models
Quantized. Fine-tuned. Air-gapped.

Every model is GGUF-quantized, open-source, and deployable entirely within your perimeter. No data leaves your infrastructure. Available now on Hugging Face.

Most Popular

Stack 4.0 Agentic

3B parameters · Q4_K_M GGUF

2.1K
downloads

Fine-tuned Qwen2.5-Coder-3B for agentic workflows. Handles function calling, tool use, and multi-turn conversations. Optimized for regulated environments.

Tool-calling agents, RAG, multi-step reasoning
my-ai-stack/Stack-4.0-Qwen-3B-Agentic
View on Hugging Face →

Stack 3.0 Omni-Nexus

8x7B MoE · Q4_0 GGUF

1.4K
downloads

Mixture-of-experts model with 8x7B active parameters. Multilingual synthesis and advanced code generation for enterprise workloads.

Complex reasoning, code generation, multilingual
my-ai-stack/Stack-3.0-Omni-Nexus
View on Hugging Face →
Training Data

Stack 4.0 Training Dataset

27K examples · Parquet / JSONL

High-quality agentic training dataset: multi-turn tool conversations, function-calling examples, and enterprise workflow patterns. Available for custom fine-tuning.

Fine-tuning, RLHF, agentic training
my-ai-stack/Stack-4.0-Dataset
View on Hugging Face →

Need a custom model for your infrastructure?

We quantize, fine-tune, and deploy custom GGUF models for enterprise VPCs. Sovereign deployment options available (air-gapped, VPC, on-premise).