What is an "open-weight" model, and how is it different from open source and proprietary models?

beginner

← All Open-Source & Local LLMs questions

Re-explain

An open-weight model ships its trained parameters as a download. You can run it, fine-tune it, and inspect the weights. What you usually do not get is the training data or the training code.

Open source, strictly speaking, also publishes the data, the code, and the recipe under a standard license. Proprietary models keep the weights in-house, so you only reach them over an API.

	Weights	Training data + code	Where it runs
Open-weight	downloadable	usually not released	your hardware, or any provider
Open source	downloadable	released	your hardware, or any provider
Proprietary	never released	not released	vendor API only

Most models people call "open source" are really open-weight. Llama weights are free to download, but the license adds use limits, so read it. Gemma moved to plain Apache 2.0 with Gemma 4.

Rewriting in plainer words…

This answer doesn't lend itself to a diagram - it reads best . No credits were charged.

Why there's no diagram: “”

The interactive diagram is below the answer - jump to diagram ↓ · Below it, the related concept . Jump to it ↓

The diagram below the answer is the concept . Jump to it ↓

Tailored explanation · switch back to · ·

What should the new diagram focus on?

Guided practice for Open-Source & Local LLMs One question at a time. Answer out loud, get graded, see what you missed.

How well did you know this?

AI:

Saved in this browser - sign in to keep your review list.

How should your speech become text?

Listening… your words appear above as you speak - tap Stop when you're done.

Recording · cr - tap Stop & transcribe when you're done.

Transcribing with AI…

Voice:

Keep going - a few more words and AI can grade it.

Interview lens How interviewers actually use this question

Likely follow-ups

What would you check in a model's license before shipping it in a commercial product? Commercial use terms, user-count or size thresholds, acceptable-use limits, attribution rules, and whether fine-tuned derivatives inherit the terms.
When would you choose an open-weight model over a proprietary API? Data that can't leave your network, cost at high steady volume, fine-tuning, pinning a version. Weigh it against GPU ops work and a possible quality gap.
You have the weights but not the training data. What can't you do because of that? You can't reproduce or audit training, check for benchmark contamination, or know exactly what the model saw. Running and fine-tuning still work.
Does open-weight mean you have to run the model on your own GPUs? No. Hosted providers serve open-weight models over an API, so you can start hosted and move to self-hosting later with the same model.
How would you adapt an open-weight model to your own domain data? Fine-tune the weights, usually with LoRA or QLoRA to cut memory, and confirm the license allows derivatives. Try prompting or RAG first.

What you can say

An open-weight model is one whose trained weights you can download, so you can run it, fine-tune it, and inspect it yourself.
What you usually don't get is the training data or the training code, so you can't rebuild the model from scratch.
Open source, strictly speaking, also releases the data, the code, and the recipe under a standard license.
A proprietary model never leaves the vendor, so the only way to use it is through their API.
Most models people call open source are really open-weight. Llama's license adds use limits, while Gemma 4 moved to plain Apache 2.0.
So I read the license before building on any open model, because downloadable weights don't mean you can use them however you like.

Weak answers to avoid

Treats open-weight and open source as the same Most so-called open source models only ship weights. Say you can run and tune them, but without the data and code you can't reproduce or audit training.
Assumes downloadable means free for any use Licenses like Llama's add use limits. Say you read the license before shipping, and contrast it with a plain permissive license like Apache 2.0.
Thinks open-weight means you must self-host Any provider can serve open weights too. The real difference is that you choose where it runs, while a proprietary model only runs behind the vendor's API.

Interview lens

Likely follow-ups, what you can say, and the weak answers to avoid.

Want a quick review of the fundamentals? See the Open-Source & Local LLMs cheatsheet.

← Back to all Open-Source & Local LLMs questions

Pro · $10/mo

48 of 55 Open-Source & Local LLMs answers are in Pro.

Full answers, code samples, and AI explanations that go simpler or deeper. Cancel anytime.

Full answers + code
AI explanations, simpler or deeper
1,000 AI credits / month
Cancel anytime

Get Pro → See plans & pricing →