Your Mac, the AI backend

Apple silicon AI power,
behind a public URL.

One page about the machine you already own: Unified Memory, the GPU it shares it with, and how UrlyUp lets others use it securely.

AI · Apple silicon

A gigantically powerful GPU, already on your desk.

Apple Silicon gives the CPU and the GPU one shared pool of Unified Memory. There is no VRAM ceiling to hit and no copies across a bus: models and workloads that demand expensive dedicated GPUs elsewhere run straight out of the memory your Mac already has. UrlyUp puts that machine behind a public URL.

✓
Core AI ready, zero rewrites
Expose your existing tools and CLIs to Apple's new Core AI framework through a Runner. No compatibility layer, no porting: if it runs on your Mac, the framework can call it.
✓
Bring your own tokens
Anthropic and OpenAI keys work natively. Paste the token into the Runner's environment; requests to either provider flow out exactly as they would from any standard client, with no proxying middleman altering the contract.
✓
Shared, securely
Anyone you authorize reaches the GPU over the same HTTPS public URLs as every other Runner. No ports opened on the Mac, no VPN to configure; the protected-space isolation model applies to AI workloads too.
✓
Built for heavy work like coding
Agents, code assistants, batch generation and local model serving get near bare-metal GPU throughput, with the Mac's memory bandwidth doing the heavy lifting instead of a rented datacenter card.
Mac · GPU pool
Serving coding agents over HTTPS
Unified memory GPU cores ✓ Secure URL live 
# provider tokens, native
ANTHROPIC_API_KEY=sk-ant-…****
OPENAI_API_KEY=sk-…****

# any authorized user, from anywhere
→ https://my-ai-on.urlyup.com  ✓ 200

One machine, many builders: your Mac's GPU serves the agents you trust, behind the same HTTPS URLs as every other Runner.