Lattice Logo
Lattice
Buy $99

Configure Infrastructure.

Model. Hardware. Framework.

Stacks define your AI infrastructure: which model to use, what hardware to run it on, and which framework to serve it with. Get AI-suggested configurations optimized for your scenario.

View docs Get Lattice
lattice.app/stacks
Lattice Stack configuration showing model provider selection, hardware options, and framework choices
Stack configuration showing model provider and variant selection

Model Selection

Choose from supported LLM providers and models. Compare capabilities, pricing, and performance characteristics to find the right fit for your use case.

  • Anthropic Claude: Haiku, Sonnet, Opus variants
  • OpenAI: GPT-4, GPT-4o, o1 reasoning models
  • Google Gemini: Pro, Flash, Ultra options
  • Meta Llama: 3.3 70B, 3.1 405B for self-hosting
  • Mistral AI: Large, 7B edge deployments
  • Embedding Models: text-embedding-3, voyage-3
Create your first stack
Hardware configuration showing GPU type and memory options

Hardware Configuration

Specify GPU requirements for self-hosted deployments. Lattice helps you understand hardware needs for different models and throughput targets.

  • NVIDIA GPUs: A100, H100, L40S options
  • Memory Calculator: VRAM requirements per model
  • Multi-GPU: Tensor parallelism configurations
  • Cloud Providers: AWS, GCP, Azure instances
Configure hardware options
Framework configuration showing inference engine options

Framework Selection

Select inference frameworks and serving solutions. From vLLM to TensorRT-LLM, choose the right tools for your deployment requirements.

  • vLLM: High-throughput serving with PagedAttention
  • TensorRT-LLM: NVIDIA-optimized inference
  • Triton Server: Multi-model serving platform
  • SGLang: Structured generation optimization
Select inference framework
AI-suggested stack recommendations with cost and performance analysis

AI-Suggested Stacks

Let Lattice recommend optimal stack configurations based on your scenario. Get suggestions grounded in vendor documentation and real-world benchmarks.

  • Scenario-Optimized: Match SLO requirements
  • Cost Analysis: Price vs performance tradeoffs
  • Benchmark Citations: Vendor performance data
  • Comparison View: Side-by-side stack options
Get AI recommendations

Technical Specifications

Everything you need to know about stack configuration options and supported infrastructure in Lattice.

Model Providers

  • API Providers - Anthropic, OpenAI, Google, Mistral
  • Self-Hosted - Meta Llama, NVIDIA NIM
  • Embeddings - OpenAI, Voyage, Cohere

Hardware Options

  • GPUs - A100, H100, L40S, RTX 4090
  • Memory - 40GB, 80GB, multi-GPU configs
  • Platforms - AWS, GCP, Azure, Lambda Labs

Frameworks

  • Serving - vLLM, TensorRT-LLM, Triton
  • Training - DeepSpeed, FSDP, Megatron
  • Optimization - FlashAttention, GPTQ, AWQ
View full documentation

Learn More About Stacks

Explore guides and documentation to configure optimal AI infrastructure.

View Stacks documentation

Stacks Documentation

Configure models, hardware, and frameworks for your AI infrastructure.

Configure Your AI Stack

Get AI-powered stack recommendations optimized for your specific requirements.

Get Lattice for $99