ChatGPT uses your Codex allowance. Claude runs through the official Claude Code app. Claude extra usage follows your own provider settings.
Connect in a few steps
Open the helper once. It installs what is needed, opens your provider’s own sign-in, and keeps your connection running.
- Download and unzip the helper for this computer.
- Open Start-ACENET in the unzipped folder. If your computer asks, allow the helper you just downloaded to run.
- Paste this code into the helper, then sign into your provider.
Waiting for your computer to connect…
The helper installs a Python runtime and the official provider app when missing, then starts your runner at login. Keep your computer awake while tasks run. Windows setup is preview and has not been tested on a Windows machine.
Open-source models
Choose a model on your computer for execution. Your connected subscription still plans and reviews the work.
Install Ollama to get started ↗
Add a model
Downloads stay on your computer. Allow extra memory beyond the download size.
Qwen3 · 4BSmaller model · about 2.5 GB
Qwen3 · 8BLarger model · about 5.2 GB
First time using a local model?
- Install and open Ollama ↗ on the connected computer.
- Choose Download & use above. ACENET handles the model download and selection.
Already using LM Studio? Turn on its local server and load a chat model. ACENET will find it automatically.
If your helper is from an earlier version, reconnect using the latest download above. Hosted API providers and Ollama cloud models are excluded from this local picker.
Other subscriptions
Gemini and other providers are not connected in this release. A subscription is supported only when its provider permits an integration and a working native runner is available.
Claude preview requires a current, unmodified Claude Code installation. Provider terms. Native Claude sign-in and a response have been verified; the full workflow remains preview.