Open Systems, Blockchain & Business Daily

OSBBD

AI, money and the software behind them.

Mistral Small 4

Mistral Small 4 is an open-weight language model from Mistral AI, released on March 16, 2026.

Developer
Mistral AI
Released
March 16, 2026
Total parameters
119B
Active parameters per token
6.5B
Architecture
MoE
Context window
256K
Input
Text, image
License
Apache 2.0
Commercial use
Yes
Official weights, GB
120.9 GB
Hugging Face repository
huggingface.co
Checked on

Mistral Small 4 is a mixture-of-experts model with 119B total parameters, of which 6.5B are active for each token. Mistral AI lists a context window of 256K tokens. The model accepts text and image input. The weights are released under Apache 2.0, which allows commercial use.

The official weights on Hugging Face take about 121 GB. Running the model needs at least that much memory across GPUs and system RAM, plus room for the context cache; quantized versions need less.

Total parameters, billions
119B
Active parameters, billions
6.5B
Context window, tokens
256K
License type
Apache 2.0
License conditions
Apache 2.0, with no user or revenue limits.
Weights precision
FP8 + BF16
Official quantized or alternative versions
mistralai/Mistral-Small-4-119B-2603-NVFP4; mistralai/Mistral-Small-4-119B-2603-eagle (speculative-decoding head)
License text
apache.org
Sources
PageURL
Hugging Face model cardhttps://huggingface.co/mistralai/Mistral-Small-4-119B-2603/raw/main/README.md
Hugging Face weights indexhttps://huggingface.co/mistralai/Mistral-Small-4-119B-2603/raw/main/model.safetensors.index.json
Hugging Face APIhttps://huggingface.co/api/models/mistralai/Mistral-Small-4-119B-2603?expand=safetensors
mistral.ai announcementhttps://mistral.ai/news/mistral-small-4/

Questions

What license does Mistral Small 4 use?
The weights are released under Apache 2.0, which allows commercial use with no user or revenue limits.
How much memory does Mistral Small 4 need?
The official weights take about 121 GB (FP8 + BF16), and all 119B parameters must be loaded even though 6.5B are active per token. Mistral AI also publishes an NVFP4 version; the memory calculator checks which setups fit.
How many parameters does Mistral Small 4 have?
119B in total, of which 6.5B are active for each token, with a context window of 256K tokens.
Is Mistral Small 4 a small model?
Not in memory terms: it is a mixture-of-experts model with 119B total parameters, though only 6.5B are active per token, which keeps it fast.

Sources

Every figure comes from the developer’s model card, license text and official announcements, and from the model’s Hugging Face repository metadata. Each model page lists its sources and the date they were checked; the register is rechecked every week.

OSBBD

What are you looking for?

Search by headline, topic or keyword.