LV

Local LLM Visual Benchmark

Browse visual benchmark outputs by prompt, model, or table.

a side quest by eeshans.com

Machine

Welcome — 5 steps to your first benchmark

1Start oMLX or LM Studio
2Load a model
3Pick a prompt
4Generate with your tool
5Capture preview

Prompt comparison

0

Browse visual benchmark outputs by prompt, model, or table.

Loading saved runs.

© 2025–2026 Eeshan Srivastava Personal project · MIT License · Non-commercial Source LinkedIn

Run detail

Prompt

0 chars

Delete run?

This removes the run folder from disk.

This cannot be undone.

The whole run folder will be deleted, including metadata, prompt, HTML, preview, and raw response files.

Run folder

Prepare run slot

Choose a prompt and model to generate a run folder.

oMLX checking LM Studio checking

Coding harness

Model source

Prompt

Discovered model

Generated prompt

Copy this into your external tool after preparing the slot.

Visual prompt

No run slot prepared yet.

Setup

Refresh oMLX or LM Studio models, then prepare runs and save generated artifacts.

Local model source

Model sources

Discover oMLX and LM Studio models. LM Studio models can still be mirrored into Pi and OpenCode when needed.

0 models

oMLX oMLX URL

Waiting for local API.

LM Studio LM Studio URL

Waiting for local API.

Config sync

Pi config Checking… ~/.pi/agent/models.json

OpenCode config Checking… ~/.config/opencode/opencode.json

Start LM Studio, refresh its model list, then sync discovered LM Studio models to your tool configs.

Model inventory

Live oMLX and LM Studio models plus saved run history.

Model Source Config status