Drop-in API compatibility
Use an existing OpenAI-compatible client against `cocore.dev/v1` with a `cocore-...` key. The site says the standard chat/completions shape, streaming, and tool calls work without code changes.
co/core
co/core is a cooperative for AI inference that pools member-owned Macs to run open models. OpenAI-compatible clients and a macOS app for contributors.
Member-owned AI inference
co/core is a cooperative for AI inference built around member-owned Macs. Instead of sending every request to a large cloud provider, it pools compute from people who already have the hardware and routes jobs to those machines to run open models for each other.
The service presents itself as an experiment in infrastructure that is built, shared, and owned together. It exposes an OpenAI-compatible API, so existing client code can point at `cocore.dev/v1` and keep using the usual chat and streaming workflow while the network handles the inference.
The public pages also emphasize transparency. Each job writes a signed receipt, the records are designed to be verifiable on their own, and the protocol is framed as an open standard that others could use to run a different exchange with different rules.
For contributors, co/core offers a macOS app for Apple Silicon that turns an idle Mac into part of the network. Members can pick models, keep the app running in the background, and receive credits for the work their machine performs, with settlement and member payouts described in public records.
What co/core does
Member-run inference network
Request jobs are routed to member-owned compute instead of a single cloud provider. The product positions this as pooled AI inference across Macs people already own.
Verifiable job receipts
Each job produces a signed, public receipt that can be verified against the published spec and signer identity. The site emphasizes that the record can be checked without calling co/core again.
Open standard and exchange model
The system describes an open standard for inference requests and post-job records. That allows a different exchange to read and write the same records with its own rules.
Mac contribution app
Members can install the macOS app, sign in, pick models, and let their Mac run jobs while they are not using it. The app is described as self-updating and not requiring Terminal use.
When people use co/core
Run familiar code against a different backend
Developers with existing OpenAI-compatible code can swap the base URL and API key, then continue using the same chat/completions and streaming flow against co/core.
Share spare Mac compute
People with an Apple Silicon Mac can install the app, choose a model or two, and let the machine contribute compute while they are not using it.
Verify what ran and who ran it
Users who want more transparency around inference can rely on signed receipts and public settlement records rather than a black-box cloud bill.
Explore an open exchange model
Anyone interested in alternative AI infrastructure can use the open records and open standard to understand how the exchange works or potentially run a different exchange with the same data model.
Pros and Cons
Pros
- Works with existing OpenAI-compatible code and clients.
- Lets members contribute Macs they already own rather than renting from a cloud provider.
- Produces signed, public receipts that can be verified against the published records and spec.
- Describes an open standard, so the exchange model is not locked to a single operator.
- Has a macOS app path for people who want to contribute compute without using Terminal.
Cons
- The pricing page in the provided source returns Not Found, so the current pricing structure is not visible here.
- The strongest setup details shown are for macOS on Apple Silicon; broader platform support is not confirmed in the supplied pages.
FAQ
What is co/core?
It is a cooperative for AI inference where members pool the Macs they already own to run open models for one another.
How do I connect existing code?
The site says you can point an existing OpenAI-compatible client at `cocore.dev/v1` and use a `cocore-...` API key. It describes streaming, tool calls, and the usual chat/completions shape as working without code changes.
What hardware does it use?
The network is designed to run open models on member-owned Macs, and the site highlights an app for macOS on Apple Silicon.
Is there pricing information?
The public pages show a free-to-start network and a pricing page that currently returns Not Found, so no paid plan details are visible in the provided source.
Who is it for?
The site describes both sides of the network: people can use the co-op's inference network, or they can install the app, sign in, pick models, and contribute their Mac's compute while they're not using it.
Quick Facts
- Category
- AI Infrastructure
- Primary model
- Cooperative AI inference
- Platform
- macOS app for Apple Silicon; API access via OpenAI-compatible clients
- Source domain
- cocore.dev
- Pricing page
- Not Found in provided source
- Notable workflow
- Point existing code at `cocore.dev/v1` or install the app to share a Mac
co/core Alternatives
AakarDev AI
AakarDev AI helps teams manage AI provider access, project setup, logs, and analytics in one dashboard. BYOK support included.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repos and verifies changes with compilers, debuggers, sanitizers, and tests.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads in Firecracker micro-VMs with private networking and SDK, CLI, or MCP control.
hob
hob is an independent workspace for coding agents, with local control over sessions, terminals, history, routing, and follow-up work.
Ably Chat
Ably Chat is a chat API platform for custom realtime chat apps, with rooms, typing indicators, presence, reactions, message updates and usage-based pricing.
Manta AI
Manta AI is an autonomous web app testing tool that maps app behavior, catches regressions, and generates tests from a URL, no scripts or selectors needed.