🖥️ LM Studio — Tool Hub: Facts, Tutorials & Verdict
Everything about LM Studio in one place: the desktop way to run LLMs locally, its VRAM-fit model browser, limits, best tutorials, and our tested guides.
Derek Holt · Local AI & Hardware Writer
· Updated · 2 min read
⚡ TL;DR — quick answers
- Is LM Studio free?
- Free to download and use as a desktop app. The app itself is proprietary (unlike open-source Ollama); check current terms for team/commercial deployment.
- What makes LM Studio different?
- The model browser shows each quant's size against your actual VRAM before downloading, and a manual GPU-offload slider gives fine control that automatic runtimes lack.
- Should I use LM Studio or Ollama?
- Pick LM Studio if you want a GUI, a model browser that shows each quant against your actual VRAM before you download, and a manual GPU-offload slider. Pick Ollama if you want a scriptable CLI and an OpenAI-compatible server to point other tools at. They are not mutually exclusive and plenty of people run both on the same machine for different jobs.
At a glance
| What it is | Desktop app for downloading, chatting with & serving local LLMs |
| Price | Free to use (proprietary app) |
| Platforms | Windows, macOS, Linux (lmstudio.ai) |
| Signature strengths | VRAM-fit model browser, manual GPU-offload slider, one-click API server |
| Weak spots | Not open-source, heavier than a CLI, less automation-friendly |
| Engine | llama.cpp under the hood — same GGUF models as Ollama |
| Best for | Beginners, model exploration, parameter tuning |
Why it's the best cockpit for local models
LM Studio made local LLMs feel like a product: search a model, see which quants fit your card before downloading, chat, tweak samplers, and flip on an OpenAI-compatible server when an app needs it. That will-it-fit UX is the single feature we point beginners at — it prevents the classic first-day mistake of downloading a model twice your VRAM.
On our 16GB RTX 4080 it's the lab bench: every new model release gets tested here first, with the offload slider rescuing borderline fits that automatic runtimes fumble. What it isn't is infrastructure — for always-on APIs and scripting, Ollama wins, which is why we run both with different jobs.
What changed — Bionic, LM Studio's agent app (July 2026)
The July 16, 2026 launch of Bionic is the biggest move in LM Studio's history: a separate, Mac-first agent app (not an update to LM Studio itself) that runs open models against your files instead of just chatting. Code projects link to a local folder for investigate/edit/debug work with inline diffs; Work projects process documents in a sandbox with automatic checkpoints for rollback; a system-wide voice keyboard transcribes offline using Mistral's Voxtral; and an optional Secure Cloud fallback handles jobs too big for local hardware under a stated zero-data-retention policy. Point releases have moved quickly — 1.0.5 landed within weeks — while classic LM Studio continues its own track (0.4.x).
We've tested and broken this down in two dedicated pages: the full Bionic guide (what it does, what it can't, the threat-model question agents raise) and Bionic vs Ollama's agent mode — because both local-AI heavyweights went agentic within a week of each other.
Official resources
- lmstudio.ai — download and docs
- LM Studio docs — local server, CLI and SDK reference
- LM Studio changelog — Bionic and 0.4.x release notes
Go deeper
- Ollama vs LM Studio — our same-hardware verdict
- Best GPU for local AI — the VRAM math behind the fit hints
- Best local AI tools — the full tested stack
Prefer video? Hand-picked walkthroughs
Reading is faster, but if you want to see it done, these are the best tutorials we vetted for this topic:
Frequently asked questions
▸Is LM Studio free?
Free to download and use as a desktop app. The app itself is proprietary (unlike open-source Ollama); check current terms for team/commercial deployment.
▸What makes LM Studio different?
The model browser shows each quant's size against your actual VRAM before downloading, and a manual GPU-offload slider gives fine control that automatic runtimes lack.
▸Should I use LM Studio or Ollama?
Pick LM Studio if you want a GUI, a model browser that shows each quant against your actual VRAM before you download, and a manual GPU-offload slider. Pick Ollama if you want a scriptable CLI and an OpenAI-compatible server to point other tools at. They are not mutually exclusive and plenty of people run both on the same machine for different jobs.
The 5 best AI video finds, every week
New models, tested prompts, and what actually worked in our production — one short email a week. No spam, unsubscribe anytime.
Written by Derek Holt
Local AI & Hardware Writer
Runs the site's local-inference rig and benchmarks every GPU, quant, and speed-stack claim on it personally before it goes in a guide. Will not shut up about VRAM bandwidth.
Explore these topics
Every guide, comparison and prompt library we have on each.
Keep learning
Local LLMs · Ollama & LM Studio
GuidesLocal LLMs on Apple Silicon: Unified Memory Wins
A 64GB MacBook loads models a 24GB RTX 4090 cannot touch. What unified memory buys you, where MLX still beats everything, and which stack to actually run.
