AI/ML Products

Open Source Models, Frameworks & Tools

Explore open source artificial intelligence and machine learning products. From LLMs to AI infrastructure, enjoy AI innovation without vendor constraints.

Book a Discovery Call Explore Nebari

Open Innovation

AI/ML Solutions Built for Ownership

Discover the tools and frameworks to perfect your AI journey. From cutting-edge models to robust infrastructure, we have everything you need to accelerate innovation. Don’t see the open source product you need? Contact us — we likely have it.

Models

Large language & vision models

A curated catalog of leading open models, ready to run inside your own environment — no vendor constraints, no data leaving your walls.

🚀 Large LLMs

Optimized for high-scale performance

Frontier-scale models for advanced use cases that demand maximum capability.

  • Llama-3.1-405B — Meta’s flagship, competitive with proprietary systems across a wide range of tasks.
  • Mistral-Large (123B) — notable for its efficiency and competitive performance.
  • Jamba 1.5 Large (398B / 94B-active) — hybrid Transformer-Mamba MoE with 256K context and high throughput.
  • DeepSeek-V2.5 (236B / 21B-active) — mixture-of-experts design activating 21B parameters per token.
  • Mixtral 8x22 (141B / 39B-active) — sparse MoE balancing performance and efficiency.
  • Grok-1 (314B / 78B-active) — designed for general-purpose tasks.
  • Command R+ (104B) — tailored to long context and multi-step agentic RAG with tool use.
  • Nemotron-4-340B — NVIDIA’s large general-purpose model with extensive alignment.

🛠️ Mid-Sized LLMs

Balanced for efficiency and performance

Strong capability in smaller, more affordable environments.

  • Llama-3.3-70B — a balance between performance and resource requirements.
  • Qwen-2.5-72B (and 32B) — efficient processing with strong performance across applications.
  • Yi-1.5-34B — optimized for multilingual support and diverse task performance.
  • Gemma-2-27B — focused on coherent, contextually relevant conversational AI.
  • Mixtral-8x7B — sparse MoE with 8 experts of 7B parameters, ~12.9B active per token.

💻 Code / Logic LLMs

Built for code generation and reasoning

Models specialized for programming languages and logical reasoning.

  • Qwen2.5-Coder-32B — code generation and understanding across many languages.
  • QwQ-32B-Preview — Qwen with Questions, focused on internal-dialog-like reasoning.
  • CodeGeeX4-All-9B — a smaller model for code generation across languages.
  • Codestral-22B — Mistral AI’s code model, surpassing Llama3 70B on HumanEval FIM.
  • CodeLlama (34B, 70B) — Meta Llama variants fine-tuned for code.
  • StarCoder2 (StarChat2)-15B — code generation plus interactive coding assistance.

🖼️ Vision Language Models

Bridging visual and textual understanding

Multimodal models that reason across images, documents, and text.

  • PaliGemma — SigLIP-So400m vision encoder paired with Gemma-2B for versatile tasks.
  • CogVLM — integrates visual and textual data for comprehensive understanding.
  • Qwen2-VL — a ViT-enhanced Qwen-VL for seamless image and video input.
  • Molmo (7B, 72B) — multimodal models for advanced vision-language understanding.
  • Pixtral-12B — leading multimodal performance on natural images and documents.
  • Phi3.5-vision — a 4.2B multimodal model efficient enough for modern smartphones.

🔍 Smaller LLMs — lightweight models for rapid deployment, such as Llama-3.1-8B, offer competitive performance in a compact package suited to a wide range of applications.

AI/ML Infrastructure

Power your models with the right tools

Frameworks engineered for performance and scalability.

JAX

High-performance machine learning with automatic differentiation.

PyTorch

A leading deep learning framework.

Ray

Distributed computing made easy.

Dask

Flexible parallel computing for large datasets.

Hugging Face

Democratizing AI development (some models require purchased licensing).

Lightning

Streamlining deep learning research (some models require purchased licensing).

Data for AI

Organize and process your data

Powerful data processing tools, databases, and vector stores to feed your models with clean, fast, reliable data.

Kafka

Real-time data streaming.

Spark

Unified analytics for large-scale data.

Flink

Stream processing at its finest.

Airflow

Workflow automation made simple.

Presto

Interactive querying at scale.

ElasticSearch

Advanced search and analytics.

Iceberg

Modern table format for data lakes.

Vector Databases

Pinecone, Weaviate, Milvus, and more — efficiently store and query embeddings.

AI Ops

Deploy, manage, and monitor with confidence

Nebari

Your foundation for scalable AI operations.

Conda

Streamlining package management and reproducible environments.

vLLM & Triton

Simplifying deployment and inference at scale.

End to End

Seamless AI Stack Integration

We make all the pieces work together. Whether you’re building pipelines, deploying models, or integrating AI solutions, our team ensures smooth implementation for your stack.

Let’s Talk

Looking for something specific or need guidance on choosing the right tools? We want to hear from you.

Talk to Us