Your Laptop Is Sitting on Untapped AI Power Right Now
→ Get the Hyperspace Pods install link at agents.hyper.space
🏄 If this feels confusing — it's time to start surfing.
You're not behind. You just haven't learned to ride the wave yet.
While everyone else is drowning in AI tools they don't understand, 900+ builders in The AI Surfer community are mastering terminals, agents, automation, and everything in between — and doing exactly this kind of thing every week.
This is what AI Commanders do: pool their hardware, cut their cloud bills, and run models on their own terms. No permission needed.
→ Join 900+ AI Commanders at TheAISurfer.com — and stop watching the wave from the shore.
Set It Up Right Now — Step by Step
No fluff. Here's everything you need to go from zero to a running AI cluster today.
💡 Don't want to follow these steps manually? Paste the entire setup guide into Claude and say: "Walk me through this step by step and run each command for me." Claude can execute terminal commands directly and guide you through the full install in a single conversation.
Step 1: Install the Hyperspace CLI
Open your terminal and paste one command. (Just copy, paste, and hit enter — your machine does the rest.)
Mac / Linux:
curl -fsSL https://agents.hyper.space/cli | bash
Windows (PowerShell — open it from the Start menu, search "PowerShell"):
irm https://agents.hyper.space/cli | iex
It auto-detects your GPU, installs the binary, and gets ready to run. Verify it worked by running these two lines:
hyperspace --version
hyperspace status
✅ You should see a version number and a status like ready. If you see an error, the Hyperspace docs have a one-step fix for each platform.
Step 2: Create Your Pod
One person on the team does this part — everyone else just joins. Run this to create the pod and generate an invite link:
hyperspace pod create "my-team"
hyperspace pod invite --role member --ttl 24h
This prints a shareable link like https://hyperspace.sh/join/hp_inv_abc123 — paste it into your group chat.
Step 3: Everyone Else Joins
Your friends paste this in their terminal (swap in the real code from your invite link):
hyperspace pod join hp_inv_abc123
Machines connect automatically. The mesh forms on its own — nobody configures anything manually.
✅ Everyone should see the other machines appear in hyperspace pod status once they join.
Step 4: Shard a Big Model Across All Machines
Check what your combined hardware can run:
hyperspace pod models --shardable
hyperspace pod resources
Then fire up a model that's bigger than any single laptop could handle:
hyperspace pod shard qwen3.5:32b
VRAM cheat sheet — pick the model that fits your combined pool:
- 🟢 16 GB total (e.g. 2 × 8 GB) — Gemma 4 (fast, lightweight tasks)
- 🟢 32 GB total (e.g. 2 × 16 GB) — Qwen 3.5 32B (code and reasoning)
- 🔵 48 GB total (e.g. 3 × 16 GB) — GLM-5 Turbo (browsing and research)
- 🔵 64 GB total (e.g. 4 × 16 GB) — Qwen 2.5 72B, Llama 3.1 70B
💡 The practical setup for most pods is three models covering different jobs: Qwen 3.5 32B for code and reasoning, GLM-5 Turbo for browsing and research, Gemma 4 for fast lightweight tasks. All running on hardware you already own.
Step 5: Get Your API Key and Endpoint
Run this to print your local OpenAI-compatible endpoint and create a key:
hyperspace pod gateway
hyperspace pod keys create --name "dev-key"
✅ You'll see something like http://localhost:8080/v1 and a key starting with pk_. Copy both — you'll need them in the next step.
Step 6: Connect to Claude Code, Codex, Cursor, or Terminal
Pick your tool below and follow the one-step connection:
Claude Code (run this in terminal — just swap in your real key):
claude --api-url http://localhost:8080/v1 --api-key pk_yourkey
💡 This is the same Claude you'd use to help you install Pods in the first place. Once your pod is running, you can point Claude Code at your own hardware instead of Anthropic's servers — so Claude helps you set it up, then runs on what you built.
OpenAI Codex CLI (same format — Codex CLI accepts any OpenAI-compatible endpoint):
OPENAI_API_KEY=pk_yourkey OPENAI_BASE_URL=http://localhost:8080/v1 codex
Cursor:
Go to Settings → Models → OpenAI API Key → paste your pk_ key, then set the base URL to http://localhost:8080/v1. Hit save. That's it.
Pure terminal — no IDE needed:
hyperspace pod infer --interactive -m qwen3.5:32b
Multi-turn chat, straight in your terminal. Works without Cursor, Claude Code, or anything else.
Test any tool with a raw curl command:
curl http://localhost:8080/v1/chat/completions \
-H "Authorization: Bearer pk_yourkey" \
-H "Content-Type: application/json" \
-d '{"model":"qwen3.5:32b","messages":[{"role":"user","content":"Hello"}]}'
No cloud account. No API billing. Your IDE talks to your pod. The pod distributes inference across everyone's machines.
No monthly bill. No data leaving your hardware. No GPU rental.
The Real-World Math
Cloud AI for a team of five:
- 💸 $500–$2,000/month
- 💸 $6,000–$24,000/year
- 💸 You own nothing at the end of it
Pods with your existing hardware:
- ✅ $0/month in API costs
- ✅ Hardware you already own
- ✅ Models that compete directly with cloud offerings — Qwen 3.5 32B is competitive on SWE-bench, GLM-5 Turbo is ranked #1 on BrowseComp for tool-calling and web research
Three friends. Three laptops. One AI cluster.
Who Should Actually Try This
- 🖥️ You're on a small dev team and your AI spend keeps creeping up
- 🖥️ You're building with LLMs and want to iterate without watching costs
- 🖥️ You want to run larger models than fit on your local machine
- 🖥️ You want your code and prompts off third-party servers entirely
- 🖥️ You've got a laptop with a decent GPU sitting mostly idle
🔒 Security Notes Before You Run This
- 🔐 Your prompts never leave your hardware — inference routes through pod members' machines, not external servers
- 🔐 Only share invite links with people you trust — pod members can see inference traffic
- 🔐 Set short TTLs on invite links (
--ttl 24hor less) — don't leave permanent invite links floating around - 🔐 Don't expose your gateway port to the open internet — keep it on localhost or a trusted local network
- 🔐 Rotate your
pk_keys if a team member leaves — runhyperspace pod keys revoke
Don't Know Terminal Yet?
No shame in it — but this is the skill gap worth closing right now. AI is increasingly operated from the command line, and the people who know how to use it have a serious edge.
→ Learn to become an AI Agent Commander at TheAISurfer.com
The AI Surfer teaches you how to actually control AI tools — terminal, agents, automation, the whole stack. Once you've got those fundamentals, Pods (and everything else) clicks instantly.
Your hardware is already on your desk. The install command is right at the top. Run it.