Command A Vision
Cohere's first multimodal model, released July 2025 for enterprise visual tasks like chart analysis, OCR and document Q&A across 6 languages.
Key Specifications
- Context window
- 128K tokens
- Max output
- 8K tokens
- Knowledge cutoff
- Released
- Parameters
- 112B
- Inputs
- Text, image
- Capabilities Show details
- Structured output Return responses in structured formats like JSON.
Hosted API pricing
| Provider | Input /1M tokens | Output /1M tokens | Cost at 10M in + 2M out | |
|---|---|---|---|---|
|
|
Open weights, no hosted price | 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 Vision needs about 66 GB of GPU memory at 4-bit with 32K context. The cheapest rental that fits is 8× RTX 3060 at about $346 a month.
| Precision | Memory | Cheapest fit (32K context) | Cost /mo |
|---|---|---|---|
| 4-bitINT4 / FP4 |
66 GB
|
$346
|
|
| 8-bitFP8 / INT8 |
115 GB
|
$634
|
|
| 16-bitFP16 / BF16 |
228 GB
|
$3,168
|
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. No guarantee of runtime support, usable performance, or that a matching quantized build exists. How we estimate costs.
Frequently Asked Questions
What is Command A Vision good for?
Visual document work: charts, diagrams, tables, OCR and image questions, in 6 languages. CC-BY-NC weights, servable on a single high-memory GPU for non-commercial use.
When is Command A Vision not a good fit?
Tool use: it has no function calling, and no audio or video input. Commercial self-hosting needs a separate licence from Cohere, and its short maximum output limits long answers.
What is the cheapest way to run Command A Vision?
No provider we track hosts it; self-hosting on the cheapest rental that fits (8x RTX 3060, $0.48 per hour) costs about $346 a month.
Can I self-host Command A Vision?
Yes. The weights are CC-BY-NC licensed. At 4-bit it needs about 66 GB of GPU memory, which starts at roughly $346 a month on the cheapest rental that fits.
More from Cohere
| Model | Context | Input /1M | Output /1M |
|---|---|---|---|
|
|
256K | No hosted price | |
|
|
128K | $0.15 | $0.60 |
|
|
8K | No hosted price | |
|
|
128K | $2.50 | $10.00 |
|
|
256K | $2.50 | $10.00 |
|
|
128K | $0.80 | $3.20 |
|
|
128K | $0.04 | $0.15 |