TokenAssemble
Check a machineStack BuilderWorkflowsCompareMethodology

Browse stack

Local AI hardware compatibility

Build a local AI setup your machine can actually run.

Choose a GPU, Mac, or mini-PC and a model. See whether it fits in VRAM or unified memory, which quantization and context work, how fast it should feel, and which runtime supports it.

FitsTightWon't fittiers, never fake tok/s

Check your hardware

No machine yet? Build a complete stack

Sourced + stamped data · computes in your browser

One connected answer

The five layers of a local AI setup

TokenAssemble connects the parts other tools leave separate. Each step opens the sourced directory behind the verdict.

  1. 01 / MEMORY

    Hardware

    GPUs and complete systems

    Browse GPUsPrebuilt systems

  2. 02 / WEIGHTS

    Model

    Architecture, size, and context

    Browse models

  3. 03 / FIT

    Quant

    Measured artifact sizes per model

    Explore model quants

  4. 04 / SERVE

    Runtime

    Support, compatibility, and ease

    Browse runtimes

  5. 05 / USE

    Software

    Apps connected to local runtimes

    Browse software

Start from the job

What are you building?

Four complete workflow templates connect software, runtime, and model picks, then grade the hardware underneath.

Local coding assistant

A private Copilot alternative: an editor assistant pointed at a model your own machine serves.

See the workflow

Private chat server

A self-hosted ChatGPT-style app for your household or team — nothing leaves the box.

See the workflow

Local transcription + summary pipeline

Voice notes and meetings to searchable text and summaries, fully offline: a local STT engine feeding a small LLM.

See the workflow

Local agent stack

An agent harness with a locally served model behind it — autonomy without an API bill.

See the workflow

Plan the machine

Need hardware too?

Answer three different purchase questions: what should I build, which machine is better, and whether buying beats cloud.

Goal + budget

Build a complete AI stack

Compatible hardware, model, quant, runtime, and software—ranked within your budget.

Build my stack
Side by side

Compare hardware

Compare GPU, Mac, and mini-PC tradeoffs with links to computed feasibility checks.

Compare machines
Buy vs rent

Local vs cloud ROI

Estimate whether a local hardware purchase breaks even for your usage and electricity cost.

Run the numbers

The answer shows its work.

Fit physics, source dates, runtime gates, assumptions, and beta labels stay visible. Speed is always a tier, never a fabricated number.

MethodologyBrowse hardware dataConnect via MCP