Open Systems, Blockchain & Business Daily

OSBBD

AI, money and the software behind them.

Command A+

Command A+ is an open-weight language model from Cohere, released on May 20, 2026.

Developer
Cohere
Released
May 20, 2026
Total parameters
218B
Active parameters per token
25B
Architecture
MoE
Context window
128K
Input
Text, image
License
Apache 2.0
Commercial use
Yes
Official weights, GB
437.5 GB
Hugging Face repository
huggingface.co
Checked on

Command A+ is a mixture-of-experts model with 218B total parameters, of which 25B are active for each token. Cohere lists a context window of 128K tokens of input, with up to 64K tokens of output. 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 438 GB in BF16. 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
218B
Active parameters, billions
25B
Context window, tokens
128K
License type
Apache 2.0
License conditions
Apache 2.0, unlike earlier Cohere open-weight releases, which were non-commercial.
Weights precision
BF16
Official quantized or alternative versions
CohereLabs/command-a-plus-05-2026-fp8; CohereLabs/command-a-plus-05-2026-w4a4
License text
apache.org
Sources
PageURL
Hugging Face model pagehttps://huggingface.co/CohereLabs/command-a-plus-05-2026-bf16
cohere.com announcementhttps://cohere.com/blog/command-a-plus
Hugging Face APIhttps://huggingface.co/api/models/CohereLabs/command-a-plus-05-2026-bf16?expand=safetensors

Questions

What license does Command A+ use?
The weights are released under the Apache 2.0 license, which allows commercial use.
Can I use Command A+ commercially?
Yes. Command A+ uses Apache 2.0, unlike earlier Cohere open-weight releases such as Command A, which were non-commercial.
How much memory does Command A+ need?
The official Command A+ weights take about 438 GB (BF16). 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 Command A+ have?
Command A+ is a mixture-of-experts model with 218B total parameters, of which 25B are active for each token. Cohere lists a context window of 128K tokens of input, with up to 64K tokens of output.

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.