We scan the machine. We ask one question.
On first launch the worker builds a device profile in under two seconds. Then it asks what else usually runs on this computer. That is the only configuration on day one.
Just this app
It can use everything
A few things
Leave about half free
A lot
Be gentle, leave most free
Your Office PC can run a 7B model at about 18 tokens per second, great for summaries and drafts. Link a second machine and the pool can take more kinds of jobs.
The number is a real benchmark on that machine, cached. The recommendation is a suggestion, never a lock.
What the pool can do together
Linking machines increases which jobs you can accept and how often a worker is awake. It does not run one giant model across laptops. v1 promises chat, summarize and translate.
| Capability | Needs | Your pool |
|---|---|---|
| Chat and summarize | Text model, about 4 GB | Yes. Office PC |
| Translate | Text model | Yes. Either machine |
| Generate images | Diffusion model, about 6 GB | Needs an extension (not v1) |
| Clean audio | Audio model, about 2 GB | Needs an extension (not v1) |
| Create courses | Text + long context | Needs an extension (not v1) |
Phone as the daily surface
Submit a job. Go to sleep. Read the answer in the morning. The work ran on a machine in the house, not in a rented model. The phone does not run the model.
Slower than ChatGPT. Free forever after $29. Private. Ready by morning.
$29 once
$29 once. Local-first. No terminal. No monthly bill. Not a subscription and not affiliated with the LocalAI open-source project.
PricingQuestions
Is this the LocalAI open-source project?
No. 1LocalAITool is a StoicSoft household product. It is not affiliated with LocalAI. We market the keyword local AI. The product name is 1LocalAITool.
Do I need a terminal?
No. The happy path is machines, jobs, and morning results. Advanced exists. It is not day one.
Is it as fast as ChatGPT?
Slower than ChatGPT. Free forever after $29. Private. Ready by morning. Speed estimates come from a real benchmark on that machine, cached, not a marketing number.
If I link two PCs, do they run one giant model together?
No. Each job runs on one awake machine that fits the model. Linking increases what kinds of jobs you can accept and how often a worker is awake. It does not invent a cluster speed.
What is the price?
$29 one-time. No monthly bill.