Verified project record

AlexsJones/llmfit

This project helps determine which large language models will run adequately on a specific computer by evaluating system RAM, CPU, and GPU resources. It scores models on memory fit, estimated speed, quality, and context, presenting recommendations through a terminal interface or command-line output.

Project overview

It addresses hardware constraint uncertainty by detecting local compute resources and matching them against a model catalog using a multi-dimensional scoring system.

Project type: Model Runtime
Deployment: Refer to project documentation
License: MIT

Best for

Developers, general users, and AI engineers who need to evaluate local hardware constraints against a catalog of large language models.
Users who prefer terminal-based interfaces or require JSON output for script and agent automation.

Key capabilities

Detects system RAM, CPU, GPU/VRAM, and backend to determine available compute resources.
Scores every model in its catalog across four dimensions: memory fit, estimated speed, quality, and context.
Ships with an interactive Terminal User Interface (TUI) as the default mode for exploring hardware specs and model scores.
Provides classic terminal commands and JSON output for scripts, agents, and automation.
Measures real tok/s and TTFT against a running provider, allowing users to share results back to the project via a PR.

Limitations and risks

Minimum hardware requirements are not documented in the available facts.
Telemetry is disabled by default, and the data boundary is external optional, indicating potential but not mandatory external communication.
While coding is not required for basic installation and operation via the TUI, utilizing CLI automation and JSON output is designed for scripts and agents.
Real-world benchmarking measures performance against a running provider, requiring an active local provider instance to function.

Getting started

Install via a package manager such as Homebrew, Scoop, MacPorts, pip, or Docker. The difficulty is documented as easy because it is available via a single command on major package managers.
Run llmfit in the terminal after installation to start the default interactive Terminal User Interface.

Alternatives and comparisons

Provides a way to run and chat with large language models locally, whereas this project focuses on scoring models against hardware constraints.
Provides LLM inference via a C/C++ implementation, contrasting with this project's focus on recommending models based on detected hardware.
Serves as a high-throughput and memory-efficient inference and serving engine for LLMs, contrasting with this project's focus on recommending models based on detected hardware.

Project comparisons

Evidence and sources

README: A terminal tool that right-sizes LLM models to your system's RAM, CPU, and GPU. Detects your hardware, scores each model across quality, speed, fit, and context dimensions, and te…
README: `llmfit bench --share` measures real tok/s on your hardware and contributes it back to the project as a PR — no `gh` CLI, no third-party account.
GitHub project description: Hundreds of models & providers. One command to find what runs on your hardware.
README: scoop install llmfit
README: brew install AlexsJones/llmfit/llmfit

Source updated at: 2026-07-16T18:02:44Z

Open original GitHub project