MiMo-V2.6-Pro
Grok 4.7
111B
- Context
- 256Ktokens
- Released
- 1y ago
- Size
- —parameters
Command A is the newest model in Cohere's Command family, its line of enterprise-focused large language models. Cohere's launch announcement leads with maximum performance at minimal compute, claiming the model matches or outperforms GPT-4o and DeepSeek-V3 on agentic enterprise tasks with much greater efficiency. It is built for real-world enterprise work such as tool use, retrieval-augmented generation, agents, and multilingual use cases.
How it compares
Level
FlagshipThe lab's most capable generally available line, priced for hard tasks.
Flagship models from other labs
Vs. your stack
Add the models you use to see how this one compares.
Sign inAdditional source evidence
- Source model
- cohere/command-a-03-2025
- Source family
- command-a
- Last updated
- 2025-03-13
- Knowledge cutoff
- 2024-06-01
- Model context
- 256K
- Model output limit
- 8K
- Open weights
- Yes
- File attachments
- No
- Reasoning
- No
- Tool calling
- Yes
- Temperature control
- Yes
Downloadable weights
Published benchmarks
| Benchmark | Score | Evaluation details | Evidence |
|---|---|---|---|
| Aider Polyglot | 12 · percent correct | 2025-03-14 | Source |
Complete source record (JSON)
{
"id": "cohere/command-a-03-2025",
"name": "Command A",
"description": "Cohere command model for multilingual enterprise agents, tools, and chat",
"family": "command-a",
"last_updated": "2025-03-13",
"knowledge": "2024-06-01",
"attachment": false,
"reasoning": false,
"tool_call": true,
"temperature": true,
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"limit": {
"context": 256000,
"output": 8000
},
"benchmarks": [
{
"name": "Aider Polyglot",
"score": 12,
"source": "https://aider.chat/docs/leaderboards/",
"metric": "percent correct",
"date": "2025-03-14"
}
],
"release_date": "2025-03-13",
"open_weights": true,
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/CohereLabs/c4ai-command-a-03-2025"
}
]
}Frequently asked questions
Privacy and data
Yes, Command A has open weights, so you can self-host it for full control of your data.
Capabilities and specs
256K tokens (256,000) is the context window of Command A, and its maximum output is 8K tokens (8,000).
No, Command A supports text only as its input modality, so it cannot read images, audio, video or PDFs.
Yes, Command A supports tool calling, also known as function calling.
No, Command A does not have a reasoning or thinking mode.
2024-06-01 is the knowledge cutoff of Command A.
Quality and benchmarks
OLMo 2 32B by Ai2, olmOCR by Ai2 and Llama 3.3 Nemotron Super 49B v1 by NVIDIA are the closest alternatives to Command A, matched on model kind and release date (no comparable price).
OLMo 2 32BCommand A vs OLMo 2 32BolmOCRCommand A vs olmOCRLlama 3.3 Nemotron Super 49B v1Command A vs Llama 3.3 Nemotron Super 49B v1
Licensing and access
Command A is open weights.
Basics
Cohere made Command A, and it was released on March 13, 2025.
Command A is open weights, so you can self-host it yourself; no pricing information is listed for it.
Generated from Vertical Tabs data, updated November 2, 2026.
Discussion
No comments yet
Be the first to share your thoughts.