Match open base model to its license consideration

Attempt it

Llama 3 / 3.1 / 3.3

Mistral 7B / Mixtral

Qwen 2 / 2.5

DeepSeek-V2 / V3

Gemma 2 / 3

Quick reference

TL;DR

Open base model licenses split three ways: standard OSI licenses (Apache 2.0, MIT) and custom vendor licenses that add clauses like Meta's user threshold or Google's use-case exclusions.

Memory aid

Sign in for the mnemonic that makes this stick. See Pro pricing.

Easy to grasp

Imagine borrowing tools from five workshops to build something you'll sell. Some workshops hand over the tool and say 'use it however you like, just keep our name on the box': that's Apache and MIT. One workshop says 'free for almost everyone, but if your shop gets enormous, come back and sign a deal': that's Meta's Llama rule about huge user counts. Another says 'free, but here's a list of things you're not allowed to build': that's Google's Gemma policy. The tool works the same in every case. The difference is the paperwork attached, and that paperwork decides whether your business can ship the thing you built.

Key concepts

Open-weight models Model licensing Apache 2.0 / MIT Commercial-use compliance

Quick scan · 2–4 min

Everything important, quickly.

4 min: license propagation to derivatives + the three buckets (Apache, MIT, vendor) + Llama threshold clause + Gemma use-case exclusions + open weights versus open source + variant-specific exceptions.

Side-by-side

Model family	License	Key condition to check
Llama 3 / 3.1 / 3.3	Meta Llama Community License	Separate Meta license required above 700M monthly active users
Mistral 7B / Mixtral	Apache 2.0	Broad commercial use; keep the notice and attribution
Qwen 2 / 2.5	Apache 2.0 (most sizes)	A few variants carry their own license; check per size
DeepSeek-V2 / V3	MIT (base weights)	Model-card commercial clarifications; among the most permissive
Gemma 2 / 3	Gemma terms + Prohibited Use Policy	Categorical use-case exclusions you must review

Real-world examples

Real products, models, and research that use this idea.

Together.ai and Fireworks host Llama 4 and Mistral fine-tunes, surfacing each base model's license in the deployment console so customers see the Meta user-count clause before serving.
DeepSeek V4 shipped under MIT for its base weights, which is why many startups pick it as a fine-tuning base with minimal legal review.
Hugging Face gates Gemma and Llama downloads behind a license-acceptance click, while Apache-licensed Qwen and Mistral weights download without that gate.
Enterprises building consumer apps that may exceed 700M users often default to Apache 2.0 bases like Mistral specifically to sidestep the Llama threshold negotiation.

Interviewer follow-ups

What an interviewer would ask next. Try answering before peeking at the approach.

QHow does a base model's license propagate to a fine-tuned derivative you redistribute?

Sign in or go Pro to see the approach for this follow-up.

QWhy is open weights not the same thing as open source, and why does the distinction matter commercially?

Sign in or go Pro to see the approach for this follow-up.

QIf you fine-tune on outputs from a frontier API, how does that interact with the base weight license?

Sign in or go Pro to see the approach for this follow-up.

60-second cheat sheet

The night-before-the-interview bullets. Scan these on the way to the call.

Sign in for 5 unlocks/day or go Pro for unlimited revision sheets. See Pro pricing.

References

Primary sources. Skim if you want the original framing.