Open Systems, Blockchain & Business Daily

OSBBD

AI, money and the software behind them.

Devstral 2

Devstral 2 is an open-weight language model from Mistral AI, released on December 9, 2025.

Developer
Mistral AI
Released
December 9, 2025
Total parameters
123B
Architecture
Dense
Context window
256K
Input
Text
License
Modified MIT License
Commercial use
Yes, with conditions
Official weights, GB
128.2 GB
Hugging Face repository
huggingface.co
Checked on

Devstral 2 is a dense model with 123B parameters. Mistral AI lists a context window of 256K tokens. The model takes text input. The weights are released under the Modified MIT License. Commercial use is allowed with conditions. Companies with global consolidated monthly revenue above US$20 million need a separate commercial license from Mistral AI.

The official weights on Hugging Face take about 128 GB in FP8. 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
123B
Context window, tokens
256K
License type
Modified MIT
License conditions
Companies with global consolidated monthly revenue above US$20 million need a separate commercial license from Mistral AI.
Weights precision
FP8
License text
huggingface.co
Sources
PageURL
Hugging Face model cardhttps://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512/raw/main/README.md
License text on Hugging Facehttps://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512/raw/main/LICENSE
Hugging Face weights indexhttps://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512/raw/main/model.safetensors.index.json
mistral.ai announcementhttps://mistral.ai/news/devstral-2-vibe-cli

Questions

What license does Devstral 2 use?
The weights are released under the Modified MIT License.
Can I use Devstral 2 commercially?
Yes, with conditions. Companies with global consolidated monthly revenue above US$20 million need a separate commercial license from Mistral AI.
How much memory does Devstral 2 need?
The official Devstral 2 weights take about 128 GB (FP8). Running the model needs at least that much memory across GPUs and system RAM, plus room for the context cache; quantized versions need less. The memory calculator checks which setups fit.
How many parameters does Devstral 2 have?
Devstral 2 is a dense model with 123B parameters. Mistral AI lists a context window of 256K tokens. The smaller Devstral Small 2 has 24B.

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.