Deploy the right open-source LLM for your specific use case. From general-purpose to specialized models, all with complete data sovereignty.
Recommended: Based on your selection, we recommend Llama 4 Scout for balanced performance and cost.
Versatile models for a wide range of enterprise applications
Meta
Mistral AI
DeepSeek AI
Purpose-built models optimized for specific enterprise needs
State-of-the-art models for software development and technical documentation
Qwen3
Best coding benchmarks, 128K context
DeepSeek V3 / R1
Specialized for code completion
CodeLlama
Meta's code-focused variant
Models optimized for retrieval-augmented generation and document understanding
Command A
Built-in citations, 128K context
Llama 4 Scout
Excellent for general RAG
Mistral Small 3 24B
Efficient multilingual RAG
HIPAA-compliant models trained on medical literature and clinical data
BioMistral
Medical-tuned Mistral variant
Meditron
Clinical decision support
OpenBioLLM
Biomedical research
Models optimized for financial analysis, risk assessment, and compliance
FinGPT
Financial analysis specialist
Llama 4 Fine-tuned
Custom financial models
Qwen3 Finance
Quantitative analysis
Detailed comparison of all available models to help you make the right choice
| Model | Parameters | Context | License | Best For | GPU Memory | Monthly Cost |
|---|---|---|---|---|---|---|
|
L4
Llama 4 Meta |
109B / 400B (17B active) | 128K | Custom | General Purpose | 16GB / 140GB / 810GB | Contact for Pricing |
|
M
Mistral Mistral AI |
24B / 123B | 32K / 64K | Apache 2.0 | EU Compliance | 24GB / 80GB | Contact for Pricing |
|
Q
Qwen3 Alibaba |
8B / 32B / 235B | 128K | Apache 2.0 | Code Generation | 16GB / 65GB / 145GB | Contact for Pricing |
|
A
Command A Cohere |
111B | 128K | CC-BY-NC* | RAG & Citations | 208GB | Contact for Pricing |
|
DS
DeepSeek V3 / R1 DeepSeek |
671B (37B active) | 128K | Custom Open | Reasoning | 80GB | Contact for Pricing |
* Commercial license available separately. Costs are estimated based on typical infrastructure requirements.
Key questions to guide your model selection
General Purpose AI
→ Llama 4 or Mistral
Code Generation
→ Qwen3 or DeepSeek V3 / R1
Document Analysis
→ Command A or Llama 4
Specialized Domain
→ Domain-specific models
European Union
→ Mistral (GDPR-native)
United States
→ Llama 4 (most popular)
Asia Pacific
→ Qwen3 (strong APAC support)
Global Multi-region
→ Multiple model strategy
Small Teams
→ Compact models (Mistral Small 3 24B, Qwen3 8B)
Medium Teams
→ Mid-size models (Llama 4 Scout, Qwen3 32B)
Enterprise Teams
→ Large models (Command A, Llama 4 Maverick)
Our experts will help you choose and deploy the perfect model for your needs