Getting started

Your first local benchmark.

Set up the client, connect a model runtime and measure your own machine.

Open Aetherion Studio

Download the Windows executable. Sign in with your Aetherion account and active license. New accounts require a signed license key.

Connect your model runtime

Start Ollama and install a model that fits your memory. Refresh the model list in the client. The Windows download uses Ollama. For GGUF, use a source installation with llama-cpp-python or a build that bundles it, then choose your model folder.

Run one workload

Select an installed model and a task category. Start the benchmark in the client, then review latency, throughput and task validation together.

Keep the comparison useful

Save the JSON report. Compare runs using the same category and settings; model speed changes with hardware, context length and runtime configuration.

Detailed setup

Runtime options, troubleshooting and command-line usage.

Read the setup guide

Benchmark criteria

Explore the prompts and what each output check validates.

Explore the five tasks

Help with your setup

Tell us about your hardware, installed models and workload.

Prepare an inquiry