# CORe Winds Down Cloud Models for a Local-First AI Platform

Published 2026-11-02 · AI.info Team · AI.info

> CORe announced on November 1 that it is winding down development of its cloud models, including Genesis, and shifting to an open-source platform that runs on users’ own machines. Chats, memories and settings will stay local unless users configure an outside provider, while existing cloud models remain available during the transition.

[CORe](https://www.opencore.one/articles/going-local-first) announced on November 1 that it is winding down development of its cloud AI models, including Genesis, and rebuilding its platform to run locally on users’ machines. The next version will be open-source, work offline and keep chats, memories and settings on the user’s device unless an outside provider is configured. Existing cloud models will remain available during the transition.

The move changes CORe’s focus from making models to packaging local AI for users. The platform will keep its current interface and memory system while adding tools to find and set up local runtimes and models.

## CORe shifts from Genesis to local models

CORe describes the decision as a response to both a privacy concern and frustration with cloud AI subscriptions and per-token charges. Its announcement argues that conversations and stored memories can reveal personal details, and that keeping data on a user’s machine reduces the amount shared with a provider.

That case rests partly on an internal experiment the company says it conducted with employees’ explicit consent. CORe built a tool that generated a digital profile from chats and stored memories. For one employee who had barely used the platform, a handful of casual conversations was enough for the tool to infer the person’s state, favorite teams and occupation, the company says. CORe showed participants their profiles, then deleted the tool and its output the same day.

The experiment is presented as the company’s reason for the shift, not as evidence that cloud providers have built comparable profiles of their users. CORe makes a broader argument that data submitted to cloud AI can serve as material for profiling; its announcement does not offer an audit of other providers’ systems.

## The planned platform keeps the familiar interface

Under the plan, CORe provides the interface while users’ computers run the models through a runtime of their choice. “The platform is the interface; your computer is the datacenter,” the company writes. The platform will work offline, require no account and carry no usage metering or API bill when used locally.

Users can also configure an outside provider, in which case information may be sent beyond their machine. That exception matters: local processing is the default CORe describes, but the platform will not be limited to local models. The announcement does not name required hardware, supported runtimes or the open-source license.

CORe plans to move API configuration into the admin page and add a local AI marketplace for discovering and setting up runtimes and models. Those features are part of the announced next platform; the post does not give a launch date or say that the new version is already available.

## Cloud models stay during the transition

CORe is winding down new cloud-model development, rather than immediately removing Genesis and the rest of its existing lineup. The current models will stay available while the transition proceeds, and the company says it will provide notice before making changes.

The announcement leaves the timetable and practical requirements for the local platform unspecified. For now, the clearest commitment is the transition itself: CORe is shifting its development effort to local AI while keeping its existing cloud models available in the meantime.

## Sources

1. Original source [CORe](https://www.opencore.one/articles/going-local-first) opencore.one

## Related stories

- [NVIDIA Launches RTX Spark Laptops for Local AI and Gaming](https://ai.info/news/nvidia-launches-rtx-spark-laptops-local-ai-gaming)
- [Liquid AI Releases LFM2.5, a 24B Model for Local Agents](https://ai.info/news/liquid-ai-lfm25-24b-local-agents)
- [NVIDIA Expands RTX Spark Software Ahead of Oct. 16 Launch](https://ai.info/news/nvidia-expands-rtx-spark-software)
- [Microsoft Opens Preorders for $5,999 Surface RTX Spark Dev Box](https://ai.info/news/microsoft-surface-rtx-spark-dev-box-preorders)
- [Edge0 Serves a 35B MoE at 20 Tokens a Second in 3 GiB](https://ai.info/news/edge0-serves-35b-moe-at-20-tokens-per-second-in-3-gib)

## Explore

[More articles](https://ai.info/news)

---

Canonical: https://ai.info/news/core-winds-down-cloud-models-local-first-platform
