Free core. Your choice of compute.

Heliox core

The desktop application, Python daemon, declared action system, local records, plugin framework, and public source are available under the MIT license. Official installers do not require a Heliox subscription.

Local language models

Ollama can run a model on your own computer without per-request API charges. You supply storage, memory, electricity, and suitable CPU/GPU capacity. Model weights are separate downloads.

Cloud model providers

OpenAI, Anthropic, Gemini, or another configured provider may charge for API usage under its own current pricing. Heliox does not add a fee, but it also does not pay those charges for you.

• OpenAI API pricing
• Anthropic pricing
• Gemini API pricing

Voice models

Kokoro and Pocket TTS run locally and do not need a speech API key. Their packages and model weights consume disk, memory, CPU time, and download bandwidth. If unavailable, Heliox can fall back to the operating-system voice.

Camera, microphone, and neural work

Text workflows need no camera, microphone, or EEG device. Voice, gaze, and gesture need compatible hardware. Synthetic and recorded-EEG development needs no headset; live EEG research requires separately purchased compatible equipment.

Hosted Heliox service

No paid hosted Heliox plan or price has been announced. If one is introduced later, this page must state the price, included usage, data path, and relationship to the free MIT-licensed core.