Qwen/operational
Qwen: Qwen2.5 VL 72B Instruct
Qwen2.5-VL is proficient in recognizing common objects such as flowers, birds, fish, and insects. It is also highly capable of analyzing texts, charts, icons, graphics, and layouts within images.
128K
Context window
tokens
115K
Max output
tokens per reply
106ms
Time to first token
p50, EU region
$1.00
Output price
per 1M tokens
Fit
Where this model earns its place
vision
Image input — screenshots, scanned documents, charts, and photographed forms, without a separate OCR step.
Data handling
Requests are processed in the EU. Prompts are not used for training, and logs follow the retention period in your contract — zero-retention is available on request.
Specifications
Model
- Model ID
- qwen/qwen2.5-vl-72b-instruct
- Provider
- Qwen
- Context length
- 128K
- Max output
- 115K
- Modality
- text + image
- Licence
- Open Weight
Commercial
- Input
- $0.80
- Output
- $1.00
- Cached input
- $0.40
- Deployment
- serverless
- Availability
- operational
- Last updated
- 2026-10-26
Cost
What a month actually costs
| Workload | Input tokens | Output tokens | Monthly cost |
|---|---|---|---|
| Prototype | 1.0M | 250K | $1.05 |
| Production | 50.0M | 12.0M | $52.00 |
| At scale | 500.0M | 120.0M | $520.00 |
Illustrative volumes at list price, excluding cached-input discounts and any committed-use agreement. Your invoice shows the same rates, itemised per key.
Cheaper by caching
API
Same SDK, different base URL
cURL
curl https://api.itciphtech.com/v1/chat/completions \
-H "Authorization: Bearer $CIPHER_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen/qwen2.5-vl-72b-instruct",
"messages": [{"role": "user", "content": "Hello!"}]
}'TypeScript
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.itciphtech.com/v1",
apiKey: process.env.CIPHER_API_KEY,
});
const res = await client.chat.completions.create({
model: "qwen/qwen2.5-vl-72b-instruct",
messages: [{ role: "user", content: "Hello!" }],
});Alternatives
Close substitutes in the catalog
Need this model on dedicated hardware?
Run it single-tenant, inside your own network boundary, with the operations handled by us.