Model / Mistral AI

Mistral Medium 3.5

Trust score 64 of 100. No, open weights with custom rules.

Last checked Changed Oct 19, 2026 History

Short answer (as of Oct 29, 2026): Mistral Medium 3.5 is a local AI model from Mistral AI. Truly open? No, open weights with custom rules. Business use: yes, with guardrails. Our estimate says it needs about 94 GB of RAM. Trust score 64 of 100.

As of Oct 29, 2026: license Mistral Modified MIT License; business use yes, with guardrails; sign-up not needed; about 94 GB RAM (estimate); trust score 64/100. Sources.

Mistral's big all-in-one model for chat, reasoning, and coding. It reads images. Big companies need a paid deal.

Run it in LM StudioOfficial model pageSee sources

Trust score

64 / 100

Source, checked Oct 29, 2026

Will it say no?

Not sure yet

What this means

Truly open?

No, open weights with custom rules

Source, checked Oct 29, 2026

OK for business?

Yes, with guardrails

Source, checked Oct 29, 2026

Sign-up needed?

No

Source, checked Oct 29, 2026

Size

127.7 billion

Source, checked Oct 29, 2026

Memory window

256K tokens

Source, checked Oct 29, 2026

In non-nerd speak

Who makes it: Mistral AI. The files come from the maker's own Hugging Face account.

License: Mistral Modified MIT License. Mistral's own version of the MIT license. You can use, change, and share it, even to make money, unless your company is big.

Business use: Only if your company (or your employer) made 20 million US dollars or less last month. Bigger companies need a paid deal with Mistral.

Training data: We have not found a full public data release. Not yet checked in depth.

Will it say no?

Not sure yet. The maker's card does not say how often it says no. So we won't guess.

This comes from what the maker wrote, checked Oct 19, 2026. We have not run our own tests yet. It sits next to the trust score and does not change it. How we pick these labels.

Will it run on my computer?

RAM needed

About 94 GB

Estimate, our method

4-bit file size

About 76.6 GB

Estimate, our method

Quality tier

Top home tier

Size-based, our method

Pick a computer with at least 128 GB of memory. On a Mac, that is the unified memory. On a PC, a graphics card with that much VRAM is fastest. A PC with only normal RAM will also work, just slower.

Not sure? Use our pick-by-computer guide.

How to run it in LM Studio

  1. Get LM Studio from lmstudio.ai and open it.
  2. Open the Discover tab. Search for unsloth/Mistral-Medium-3.5-128B-GGUF.
  3. Pick a file that ends in Q4_K_M. It is the best mix of size and quality for most people.
  4. Click Download. When it is done, open a new chat and load the model.

The files are GGUF builds that LM Studio can load. We checked that this repo exists on Oct 29, 2026. Full LM Studio guide.

How to use it with Hermes

Does it fit? Yes. Its memory window is big enough for Hermes.

  1. Load the model in LM Studio. Set the context length to at least 64000.
  2. In LM Studio, open the Developer tab and start the server.
  3. In a terminal, run hermes model. Choose LM Studio and press Enter to keep http://localhost:1234/v1.
  4. Pick this model from the list. Then run hermes to start.

Steps based on the Hermes docs. Full Hermes guide.

Common questions

Is Mistral Medium 3.5 truly open source?

No, open weights with custom rules. Mistral's own version of the MIT license. You can use, change, and share it, even to make money, unless your company is big.

Can I use Mistral Medium 3.5 for my business?

Only if your company (or your employer) made 20 million US dollars or less last month. Bigger companies need a paid deal with Mistral.

Do I need to sign up to download Mistral Medium 3.5?

No. Download without an account.

How much memory does Mistral Medium 3.5 need?

Our estimate is about 94 GB of RAM for a 4-bit file of about 76.6 GB. This is a rough guide, not a test.

Change history

All changes on the site

Sources we checked

FactWhat we foundStatus
MakerMistral AIVerified Oct 29, 2026
LicenseMistral Modified MIT LicenseVerified Oct 29, 2026
License plainMistral's own version of the MIT license. You can use, change, and share it, even to make money, unless your company is big.Verified Oct 29, 2026
Truly openNo, open weights with custom rulesVerified Oct 29, 2026
Business useOnly if your company (or your employer) made 20 million US dollars or less last month. Bigger companies need a paid deal with Mistral.Verified Oct 29, 2026
Signup neededNo. Download without an account.Verified Oct 29, 2026
Parameters127.7 billionVerified Oct 29, 2026
Open training dataNot yet checkedWe have not found a full public training data release. Not yet checked in depth.
Context window256K tokensVerified Oct 29, 2026
First posted2026-03-31Verified Oct 29, 2026
Gguf filesunsloth/Mistral-Medium-3.5-128B-GGUFVerified Oct 29, 2026
Hermes okYesVerified Oct 29, 2026

If the maker's page and this page disagree, trust the maker. Tell us what we got wrong.

Newsletter

Local AI news for non-coders

Which new models are worth your time, which ones are just hype, and the license gotchas before they bite. All the useful bits, none of the nerd-party small talk.

This opens our sign-up page on Beehiiv, our email provider. Confirm your email there to join.