Supported LLM Providers
Choose the perfect model for every job - whether you need API access, open-source flexibility, or custom-trained solutions that fit your exact requirements.
Model Categories We Support
Picking the right AI model isn't just about grabbing the latest and greatest. It's about finding what works for your specific use case, budget, and compliance requirements. Here's how we break it down:
Closed-source API Models
These are your premium options – think of them as the luxury cars of the AI world. They’re powerful, constantly updated, and you pay for what you use.
- OpenAI: o3, o4-mini, o3-mini, gpt-4.1, gpt-4.5
- Anthropic: claude-opus-4, claude-sonnet-4, claude-3-7-sonnet, claude-3-5-haiku
- Mistral: Mistral Medium 3, Magistral Medium, Codestral 2, Ministral
Open-source / Self-hosted LLMs
Want more control over your destiny? These models let you run everything in-house. Perfect when you need to keep data locked down or want predictable costs at scale.
- Meta: LLaMA 3 (8B, 70B)
- Mistral: Mistral 7B, Mixtral 8x7B
- TII: Falcon 180B
- Others: Zephyr, OpenChat, Orca, Nous
Fine-Tuned Domain-Specific Models
Sometimes, you need a specialist, not a generalist. We work with models trained specifically for healthcare conversations, financial analysis, and legal document review – whatever your industry demands.
- Healthcare, finance, legal, and client-provided foundation models
- Privacy-hardened, task-specific variants built for heavily regulated environments
Model Selection Framework
Here's the thing - there's no one-size-fits-all solution. We help you navigate the trade-offs by looking at what matters for your business:
Task Type
Are you classifying support tickets, generating marketing copy, summarizing research, or building RAG systems?
Performance Requirements
Do you require sub-200ms responses, or can you process batches overnight?
Cost Sensitivity
Sometimes, the cheapest API option costs more than running your own GPU farm.
Compliance
HIPAA, SOC2, GDPR - we know which models play nice with which regulations.
Modalities
Text-only, vision processing, multilingual support, code generation, embeddings.
Model Routing & Orchestration
The real magic happens when you stop thinking about individual models and start thinking about intelligent routing. Our architecture lets you:
- Set up automatic failover when your primary model hits rate limits.
- Route different types of requests to the most cost-effective option.
- Run multi-agent workflows where different models handle what they’re best at.
- Implement confidence scoring and fallback logic to ensure you always obtain reliable results.
Framework Compatibility
We don't lock you into proprietary tooling. Everything we build works seamlessly with the frameworks your team already knows:
LangChain / LangGraph
AutoGen / CrewAI / Semantic Kernel
LlamaIndex, DSPy, Pydantic-AI
Full observability through Langfuse, PromptLayer, and OpenTelemetry
Deployment Flexibility
Whether you’re comfortable with cloud APIs or need everything running behind your firewall, we’ve got you covered:
- API access via OpenAI, Azure, Google Cloud – simple and fast
- Self-managed cloud deployment with Docker/Kubernetes for better control
- Fully isolated on-premise setup with open-weight models when security is non-negotiable