Text Generation (Vision)

Vision models behave similarly to text models, but they can accept and interpret images as well.

Model Spotlight: Qwen2.5-VL-7B-Instruct (Vision + Language)

Qwen2.5-VL-7B-Instruct is a powerful multimodal model designed to understand and process both visual and textual inputs. Whether you're building visual assistants, image-based chatbots, or AI tools that analyze pictures, this model is a seamless plug-and-play solution.

---

Getting Started with Vision-Language Inference

1. Log in to your Sentry Block account.

2. Make sure you have an active credit balance.

3. Go to the "Serverless Endpoints" tab and select the Qwen2.5-VL-7B-Instruct model.

4. Configure your parameters, upload your image (if applicable), enter your prompt, and hit Enter.

---

Additional Parameter (for Vision-Language models)

In addition to the standard generation parameters, vision-language models include:

• Image – The image URL or file to be analyzed alongside the text prompt.

---

Using the API Directly

You can integrate the model directly into your own projects via the Sentry Block API. Below are some quick examples to get started.

Important: Don’t forget to include your API key. You can find authentication details in the [API Reference].

---

cURL Example

bashCopyEditcurl -X POST "https://inference.sentryblock.com/v1/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $SENTRY_API_KEY" \
--data-raw '{
  "messages": [{
    "role": "user",
    "content": [
      {"type": "image_url", "image_url": {"url": "https://example.com/demo.jpeg"}},
      {"type": "text", "text": "What is this image?"}
    ]
  }],
  "model": "Qwen/Qwen2.5-VL-7B-Instruct",
  "max_tokens": null,
  "temperature": 1,
  "top_p": 0.9,
  "stream": false
}'

---

Python Example

pythonCopyEditimport requests
import os

url = "https://inference.sentryblock.com/v1/chat/completions"
headers = {
    "Content-Type": "application/json",
    "Authorization": f"Bearer {os.environ.get('SENTRY_API_KEY')}"
}
data = {
    "messages": [{
        "role": "user",
        "content": [
            {"type": "image_url", "image_url": {"url": "https://example.com/demo.jpeg"}},
            {"type": "text", "text": "What is this image?"}
        ]
    }],
    "model": "Qwen/Qwen2.5-VL-7B-Instruct",
    "max_tokens": None,
    "temperature": 1,
    "top_p": 0.9,
    "stream": False
}

response = requests.post(url, headers=headers, json=data)
print(response.json())

---

JavaScript Example

javascriptCopyEditconst url = "https://inference.sentryblock.com/v1/chat/completions";

const headers = {
  "Content-Type": "application/json",
  "Authorization": `Bearer ${process.env.SENTRY_API_KEY}`
};

const data = {
  messages: [{
    role: "user",
    content: [
      { type: "image_url", image_url: { url: "https://example.com/demo.jpeg" } },
      { type: "text", text: "What is this image?" }
    ]
  }],
  model: "Qwen/Qwen2.5-VL-7B-Instruct",
  max_tokens: null,
  temperature: 1,
  top_p: 0.9,
  stream: false
};

fetch(url, {
  method: 'POST',
  headers: headers,
  body: JSON.stringify(data)
})
  .then(response => response.json())
  .then(data => {
    console.log(JSON.stringify(data, null, 2));
  })
  .catch(error => console.error('Error:', error));

---

Model Identifier

Use the following model ID in your API calls:

• Qwen2.5-VL-7B-Instruct → Qwen/Qwen2.5-VL-7B-Instruct

---

Response Example (Non-Streaming)

jsonCopyEdit{
  "id": "chatcmpl-7ba48f119a564f4ea02b6a41386a3e40",
  "created": 1740689977,
  "model": "Qwen/Qwen2.5-VL-7B-Instruct",
  "object": "chat.completion",
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "This image shows ....",
        "role": "assistant"
      }
    }
  ],
  "usage": {
    "completion_tokens": 91,
    "prompt_tokens": 3604,
    "total_tokens": 3695
  }
}

---

Response Example (Streaming Enabled)

jsonCopyEdit{
  "id": "chatcmpl-3812731562554b23a32dd80fbb7d0d09",
  "created": 1740692435,
  "model": "Qwen/Qwen2.5-VL-7B-Instruct",
  "object": "chat.completion.chunk",
  "choices": [
    {
      "index": 0,
      "delta": {
        "content": " setting"
      }
    }
  ]
}

Multiple chunks like this will be streamed in real time, forming the full generated response as data is received.