A.E.V.A. — Autonomous Extended Volitional Agent

Aevonix Research / Project

Autonomous Extended Volitional Agent.

A.E.V.A. is our research into voice, vision, embodiment and out-of-model-weights recursive learning.

In development

01 / The system

How A.E.V.A. runs.

Local models handle language, vision and speech. Hermes coordinates the work; Protagine carries memory and context between tasks.

Functional overviewSelect a component

02 / Hardware

Tap a marker or choose a component to explore its hardware and role.

Model assignments reflect the September 2026 research configuration. Installed hardware can be allocated across several workloads.

03 / Models & software

Recorded benchmarks and service observations. Open a model for its hardware, test conditions and measurement history.

Reading these measurements

TPS · Tokens per second: Single-request generation speed and aggregate throughput across concurrent requests are labelled separately.
TTFT · Time to first token: Time from sending a request to receiving its first token. Context length, caching and concurrent work affect the result.
Speech timing: The first audio byte marks the start of a response stream, before speaker playback. Generation time is shown with the audio duration where recorded.
Historical records: Model versions, hardware and test conditions can change between records. Missing timings or sample counts are left unreported.