What is Command A good for?

RAG, agents and multi-step tool use in 23 languages. Weights are open under CC-BY-NC, which allows non-commercial self-hosting on a single high-memory GPU.

When is Command A not a good fit?

Image input: it is text only. Commercial self-hosting needs a separate licence from Cohere, it sits at the top of Cohere's price range, and its short maximum output limits long answers.

What is the cheapest way to run Command A?

Hosted, unless you push serious volume. Cohere charges $2.50 in / $10.00 out per 1M tokens. The cheapest rental that fits is 8x RTX 3060 at $346 a month, which costs the same as about 92M tokens a month on that API.

Can I self-host Command A?

Yes. The weights are CC-BY-NC licensed. At 4-bit it needs about 65 GB of GPU memory, which starts at roughly $346 a month on the cheapest rental that fits.

Open weights · CC-BY-NC

Command A

Cohere's 111B flagship enterprise model, released March 2025 and optimized for RAG and agentic workflows across 23 languages on two GPUs. Weights under CC-BY-NC.

Jump to: Specs Pricing Self-hosting Similar models FAQ More from Cohere

Visit website Compare vs other LLMs →

List price via Cohere

Input: $2.50 /1M tokens
Output: $10.00 /1M tokens

2 providers

Key Specifications

Context window: 256K tokens
Max output: 8K tokens
Released: Mar 2025
Parameters: 111B
Inputs: Text

Capabilities Show details: Function calling Connect to external tools, APIs, and systems. Structured output Return responses in structured formats like JSON.

Hosted API pricing

Input tokens M Output tokens M

Provider	Input /1M tokens	Output /1M tokens	Cost at 10M in + 2M out
Cohere Creator	$2.50	$10.00	$45.00	View
Azure	$2.50	$10.00	$45.00	View

Heads up: Base-tier, on-demand rates per 1M tokens; cached, batch and long-context tiers excluded. A provider may serve a shorter context or a quantized build than the creator's release. Verify before provisioning. More on how we price.

Estimated cost to self-host

Command A needs about 65 GB of GPU memory at 4-bit with 32K context. The cheapest rental that fits is 8× RTX 3060 at about $346 a month.

That costs the same as roughly 92M tokens a month on Cohere's API. Self-hosting is more expensive below that volume.

Single node ≤ 8

Precision	Memory	Cheapest fit (32K context)(Full 256K context)	Cost /mo	Break-even vs API
4-bitINT4 / FP4	65 GB80 GB	8× RTX 3060	$346$346	92M tokens /mo92M tokens /mo
8-bitFP8 / INT8	114 GB130 GB	8× RTX 4060 Ti	$634$979	169M tokens /mo261M tokens /mo
16-bitFP16 / BF16	226 GB241 GB	8× RTX A6000	$3,168$3,168	845M tokens /mo845M tokens /mo

Estimates based on median on-demand rates for Nvidia GPUs. Memory is weights plus KV cache for one request, using FP8 KV cache where supported and FP16/BF16 otherwise. Break-even assumes a 5:1 input-to-output ratio. No guarantee of runtime support, usable performance, or that a matching quantized build exists. How we estimate costs.

Similarly priced models

The models nearest Command A by blended rate, each at its own cheapest provider.

Model	Blended /1M	Input /1M	Output /1M	Context	Cutoff	vs Command A
Claude Sonnet 5 Anthropic	$3.3333	$2.00	$10.00	1M	Jan 2026	−11%
GPT-5.6 Sol OpenAI	$3.3333	$2.00	$10.00	1M	Feb 2026	−11%
Gemini 3.1 Pro Preview Google Cloud	$3.6667	$2.00	$12.00	1M	Jan 2025	−2%
GPT-5.6 Terra OpenAI	$3.6667	$2.00	$12.00	1M	Feb 2026	−2%
Kimi K3 Moonshot AI	$3.6667	$2.00	$12.00	1M		−2%
Command A This model Cohere	$3.75	$2.50	$10.00	256K
Command R+ 08-2024 Cohere	$3.75	$2.50	$10.00	128K	Feb 2023	0%
GPT-4o OpenAI	$3.75	$2.50	$10.00	128K	Oct 2023	0%
GPT-5.2 OpenAI	$3.7917	$1.75	$14.00	400K	Aug 2025	+1%
GPT-5.3-Codex OpenAI	$3.7917	$1.75	$14.00	400K	Aug 2025	+1%
GPT-5.4 OpenAI	$4.5833	$2.50	$15.00	1M	Aug 2025	+22%

Prices are USD per 1M tokens at each model's cheapest listed provider. Blended is the cost of 10M input plus 2M output tokens, spread over the 12M.

Runpod

Spin Up a GPU in Seconds - No Setup Required

Deploy AI Models Globally in 30+ Regions

Scale from 0 to 1000 GPUs on Demand

PyTorch, Tensorflow & Docker Ready

Learn more

Frequently Asked Questions

What is Command A good for?: RAG, agents and multi-step tool use in 23 languages. Weights are open under CC-BY-NC, which allows non-commercial self-hosting on a single high-memory GPU.
When is Command A not a good fit?: Image input: it is text only. Commercial self-hosting needs a separate licence from Cohere, it sits at the top of Cohere's price range, and its short maximum output limits long answers.
What is the cheapest way to run Command A?: Hosted, unless you push serious volume. Cohere charges $2.50 in / $10.00 out per 1M tokens. The cheapest rental that fits is 8x RTX 3060 at $346 a month, which costs the same as about 92M tokens a month on that API.
Can I self-host Command A?: Yes. The weights are CC-BY-NC licensed. At 4-bit it needs about 65 GB of GPU memory, which starts at roughly $346 a month on the cheapest rental that fits.

More from Cohere

Model	Context	Cutoff	Input /1M	Output /1M
Command R+ 08-2024 Open weights	128K	Feb 2023	$2.50	$10.00
Command A+ Open weights	128K		$0.80	$3.20
Command R 08-2024 Open weights	128K	Jun 2024	$0.15	$0.60
Command R7B Open weights	128K		$0.04	$0.15
Command A Reasoning Open weights	256K	Jun 2024	No hosted price
Command A Vision Open weights	128K	Jun 2024	No hosted price
Command A Translate Open weights	8K	Jun 2024	No hosted price