Command A
Pilot $4,900 fixed
- Sizes
- 111B dense
- Active params
- All parameters (dense)
- Context
- 256K
- Licence
- CC-BY-NC-4.0 · non-commercial
- Fits on
- 1× H100 80GB at 4-bit, 1× H200 141GB at FP8
- Endpoint
- llm.yourcompany.internal/v1
- Egress
- 0
- Owner
- YOU
Command Aon your own GPUs.
Command A and Command A+ on GPUs you control, with an OpenAI-compatible endpoint inside your network.
This card is the installation. Pick a size on its edge and it is re-issued.
The 72-hour pilot with Command A.
$4,900 fixed. Command A is dense, 111B parameters, 222 GB at BF16: it fits on 1× H100 80GB at 4-bit. Command A+ is a 218B mixture-of-experts model with 25B active and needs 1× H200 141GB at 4-bit.
Live on your data
72 h$4,900 fixed
Command A live on your data in 72 hours. The clock starts after access and prerequisites.
Your endpoint
/v1OpenAI-compatible
OpenAI-compatible endpoint. Your applications switch by changing the base URL.
Your hardware
- Command A1× H100 80GB at 4-bit, 1× H200 141GB at FP8
- Command A+1× H200 141GB at 4-bit, 4× H100 (320GB) at FP8
The readout
Evalon your data
Reference evaluation on your data. Readout and production scoping.
We size the hardware and source it at cost, with no markup. Ballpark: one H100 80GB ≈ $25–30k; a 4×H100 node ≈ $180–240k; smaller models run on far less.
Same SDK. Change the base URL.
The model is one parameter. Everything else in your application stays as it is.
client = OpenAI(
base_url="https://llm.yourcompany.internal/v1", # inside your network
api_key="your-key",
)
reply = client.chat.completions.create(
model="command-a",
messages=[{"role": "user", "content": "Summarise this contract."}],
)Licence, plainly.
- Command ACC-BY-NC-4.0 · non-commercialCommercial use: not permitted
- Command A+Apache 2.0Commercial use: permitted
The published Command A weights are CC-BY-NC 4.0, which does not cover commercial use. Running it in production needs a commercial licence from Cohere, arranged separately from our fee.
Command A+ is Apache 2.0, which makes it the commercial option from the same vendor.
We tell you this before the pilot, not after.
Published by Cohere. We are not affiliated with the vendor. This is a summary, not legal advice.
Related models.
- MistralMistral AI4 models, from 1× T4 16GB at 4-bitApache 2.0 / Mistral Research License (MRL) (research only)
- DeepSeek V3 / R1DeepSeek8× H100 (640GB) at 4-bitMIT (R1); DeepSeek Model Agreement (V3) (use policy (V3))
- Llama 4Meta2 models, from 1× H100 80GB at 4-bitLlama 4 Community License (conditions)
Talk to an engineer.
Tell us the model, the data and the GPUs you have. An engineer from Limassol or Auckland scopes the 72-hour pilot with you.