Coding Assistant · 200 developers
/api/size/coding-assistant-200 →- GPUs (C*)
- 34
- Winner
- h200-sxm
- Nodes
- 5
- Facility
- 46.2 kW
Generic 8× GPU HGX Node · 8 GPUs/node · 2048 GiB RAM · 15.36 TB NVMe · 400 Gbps · tok/s/user 32.17 · TTFT 163 ms · binding NONE
EchoSizer
Own the factory. Don't rent the future.
Size GPU, RAM, NVMe, network, and storage for LLM inference factories — generic catalogs, golden tests, CLI outputs in JSON · Markdown · XLSX.
Factory sizing results
Golden CLI snapshots from the Python engine (T1–T6 · 16/16).
GET /api/examples
Interactive 8B · 400 users
/api/size/interactive-8b-400 →- GPUs (C*)
- 1
- Winner
- h200-sxm
- Nodes
- 1
- Facility
- 9.2 kW
Generic 8× GPU PCIe Node · 8 GPUs/node · 1536 GiB RAM · 7.68 TB NVMe · 200 Gbps · tok/s/user 35.15 · TTFT 4 ms · binding NONE
Run the engine
pip install -e ".[dev]" python -m pytest -v echosizer run examples/factory.coding-assistant-200.yaml \ --catalogs catalogs/ \ --out-json /tmp/record.json \ --out-md /tmp/trace.md \ --out-xlsx /tmp/EchoSizer.xlsx