Aguasto Research MCP Resources
Aguasto Research offers 90 resources. A resource is data the server exposes for a client to attach as context — a file, a page, an API response — rather than an action. Each has its own page, per client.
-
API Versioning & Deprecation Policy -
Developer Sandbox & Synthetic API Tests -
Aguasto | Modular AI Infrastructure, GPU Compute & Inference -
Infrastructure -
GPU Cloud -
AI Inference -
Development & Investment -
Company -
Contact -
Privacy -
Public Research Terms of Use -
AI Compute Research Topics -
GPU systems & local clusters -
Model serving & inference -
Modular infrastructure & resources -
Compute economics & evidence -
AI Compute & Infrastructure Glossary -
Research Sources & Methodologies -
Developer API, MCP & SDK Documentation -
Aguasto Research & Resources -
GPU Library -
Model Index -
Inference Stack Library -
AI Infrastructure Engineering Guides -
AI Infrastructure Insights -
AI Infrastructure Research -
Public Infrastructure Case Analyses -
Why Modular AI Data Centers? -
Cooling & Resource Efficiency -
OEM & Strategic Collaboration -
Compare GPU & Rack References -
Compare Open-Weight Model References -
AI Infrastructure Engineering Calculators -
The Growing Friction in Hyperscale Data Center Development -
NVIDIA H100 SXM -
NVIDIA H200 SXM -
NVIDIA B200 SXM -
NVIDIA B300 -
NVIDIA GB200 NVL72 -
NVIDIA GB300 NVL72 -
NVIDIA Vera Rubin NVL72 -
AMD Instinct MI300X -
AMD Instinct MI325X -
AMD Instinct MI355X -
AMD Instinct MI455X -
AMD Helios -
GLM-4.5 -
Qwen3-235B-A22B -
DeepSeek-R1 -
Llama 3.3 70B Instruct -
Gemma 3 27B IT -
gpt-oss-120b -
Kimi-K2-Instruct -
GLM-5.3 -
Qwen3.8-27B -
DeepSeek-V4.1-Flash -
Llama 4 Scout 17B-16E Instruct -
Gemma 4 31B IT -
Kimi K3 -
vLLM -
SGLang -
TensorRT-LLM -
NVIDIA Dynamo -
NVIDIA Triton Inference Server -
AMD ROCm -
Ray Serve -
Kubernetes GPU orchestration — NVIDIA device plugin -
Model weights are only the beginning of the memory budget -
Why tokens per second is not enough to benchmark inference -
Prefill, decode and the cost of separating them -
Choosing between vLLM, SGLang and TensorRT-LLM -
From Power to Inference—Building a Distributed AI Compute Network -
Dedicated compute and shared inference: compare the whole workload -
Distributed inference reliability begins with the failure boundary -
Core Scientific: reading contracted, energized and billed capacity -
Why Modular AI Infrastructure Matters Now -
What Is a Modular AI Data Center? -
How Modular AI Data Centers Are Built and Commissioned -
Liquid Cooling for High-Density GPU Systems: Follow the Interfaces -
Why Electricity Availability Shapes AI Infrastructure Deployment -
What Is KV Cache? Capacity, Reuse and Isolation -
Continuous Batching: Throughput Without Hiding Latency -
Applied Digital: Reading the Polaris Forge 1 Delivery Milestones -
Crusoe Abilene: Separate Delivered Infrastructure from Expansion Plans -
Data Center Cooling: Choosing Air, Liquid, Dry and Hybrid Architectures -
Can Data Centers Reduce Water Consumption? Define the Water Boundary Fi… -
Measuring PUE, WUE and AI Compute Efficiency at the Right Boundary -
The Environmental Trade-offs of High-Density AI Infrastructure -
Selecting GPU Systems for Enterprise Inference -
Networking Rack-Scale AI Systems: Define the Communication Boundary
Read from the server itself, by connecting to it and calling resources/list on 5 November 2026.
A listing does not declare its resources, so asking is the only way to know them, and this
is what the server actually offers rather than what its listing claims. Names only are
stored; connect the server for each resource's type and contents.