Name: InternVL3.5 30B API: Image and Video Understanding
Brand: magicapi
SKU: api-magicapi-vision-language-multimodal-ai-lucataco-internvl3-5-30b
Availability: InStock
Author: API Market

Shashank Agarwal

# 👁️ Image and Video Understanding API (OpenGVLab InternVL3.5 30B)

## Overview
InternVL3.5 30B (InternVL3_5-30B-A3B) is an open-source multimodal large language model from OpenGVLab's InternVL3.5 family. Its developer describes it as specializing in versatile vision-language tasks with enhanced reasoning.

Send an image (`images`) or a video (`video`) with a text `prompt`, and the model answers in text: a description, an answer to a question, or step-by-step reasoning when `thinking_mode` is on. For a video, the model samples up to `max_frames` evenly spaced frames and reasons over them. The answer is returned as one text string.

## Features
- **Image understanding**: describe an image, answer questions about it or read what it shows.
- **Video understanding**: pass a `video` URL; up to `max_frames` frames (1 to 32, default 8) are extracted evenly across the clip.
- **Thinking mode**: set `thinking_mode` to `true` for step-by-step reasoning.
- **Sampling controls**: `temperature` (0 to 2, default 0.2), `top_p` (default 0.95), `top_k` (default 50), `repetition_penalty` (1 to 2, default 1) and beam search with `num_beams` (1 to 5).
- **Answer length**: `max_new_tokens` up to 8192 (default 1024).

## Use Cases
1. **Video descriptions**: summaries and detailed descriptions of short clips for search, cataloguing or accessibility.
2. **Image questions**: ask what is in a photo, screenshot or chart and get a written answer.
3. **Content tagging**: describe scenes, objects and activities to build tags or metadata.
4. **Reasoning over visuals**: turn on `thinking_mode` for questions that need several steps.

## Visual Examples
### Example Input
- `video`: [Watch the input video](https://blog.api.market/wp-content/uploads/2026/10/internvl3_5-30b-input-api-market.mp4)
- `prompt`: "Please describe this video in detail"
- Settings: `max_frames` `8`, `max_new_tokens` `1024`, `temperature` `0.2`, `top_p` `0.95`, `top_k` `50`, `num_beams` `1`, `repetition_penalty` `1`, `thinking_mode` `false`.

### Example Output
```text
In a sunlit park with lush greenery and people seated on benches, a monkey is seen riding a skateboard. The skateboard has a vibrant design with red wheels and a black deck. The monkey, with light brown fur, appears focused and balanced as it propels itself forward. The environment is lively yet relaxed, with dappled sunlight filtering through the trees, casting soft shadows on the pavement. ... The colors are warm and inviting, adding to the cheerful atmosphere of the clip.
```

## Input Parameters
| Field | Type | Required | Default | Allowed values | Description |
| --- | --- | --- | --- | --- | --- |
| `images` | string (URL) | No | none |  | Input image URL for image understanding |
| `video` | string (URL) | No | none |  | Input video file for video understanding (alternative to `images`) |
| `prompt` | string | No | `Please describe this image in detail.` |  | Text prompt or question about the image or video |
| `max_frames` | integer | No | `8` | 1 to 32 | Maximum number of frames to extract from the video (evenly spaced) |
| `max_new_tokens` | integer | No | `1024` | 1 to 8192 | Maximum number of tokens to generate |
| `temperature` | number | No | `0.2` | 0 to 2 | Temperature for sampling (higher = more random) |
| `top_p` | number | No | `0.95` | 0 to 1 | Top-p sampling parameter |
| `top_k` | integer | No | `50` | 1 or more | Top-k sampling parameter |
| `num_beams` | integer | No | `1` | 1 to 5 | Number of beams for beam search (1 = no beam search) |
| `repetition_penalty` | number | No | `1` | 1 to 2 | Repetition penalty (1.0 = no penalty) |
| `thinking_mode` | boolean | No | `false` |  | Enable thinking mode for step-by-step reasoning |

## Pricing
- **Pay As You Go (PAYG) pricing**: you only pay for what you use.
- **Price**: $0.15 per API unit.
- **1 API unit = 1 successful request with up to 96 seconds of processing.** A longer job bills one more unit for each started block of 96 seconds (for example, 100 seconds of processing is 2 units, $0.30).
- Processing time grows with `max_frames`, `max_new_tokens`, `num_beams` and `thinking_mode`.
- **Example cost**: the example request (8 frames, up to 1024 new tokens) took 25.9 seconds of processing, which is 1 unit, so $0.15.
- **Failed or cancelled predictions are not billed.**
- **GET requests to check prediction status** are free and don't count toward API unit usage.
- **No monthly subscription fee.**

## API Documentation

### Base URL
All API requests go to: `https://prod.api.market/api/v1/magicapi/vision-language-multimodal-ai-lucataco-internvl3-5-30b`

### Authentication
Send your api.market API key in the `x-api-market-key` header:
```
x-api-market-key: YOUR_API_MARKET_KEY
```

### Asynchronous Processing
**IMPORTANT**: This API is asynchronous. The flow is:
1. Send a POST request to `/predictions`. A successful request returns **201 Created** with a prediction `id` and a `status` of `starting` or `processing`.
2. Poll GET `/predictions/{id}` with that ID.
3. Keep polling while `status` is `starting` or `processing`.
4. Stop when `status` is `succeeded` (read `output`), `failed` or `canceled` (read `error`).

A polling interval of 1 second and an overall timeout of 600 seconds (10 minutes) work well.

### Endpoints

#### Create a Prediction
```bash
curl -X POST "https://prod.api.market/api/v1/magicapi/vision-language-multimodal-ai-lucataco-internvl3-5-30b/predictions" \
  -H "Content-Type: application/json" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY" \
  -d '{
    "version": "92713bb56e2e4827477501fc512d80d7287bc34e4ccb7c306afd9bdaef8b7eeb",
    "input": {
      "top_k": 50,
      "top_p": 0.95,
      "video": "https://blog.api.market/wp-content/uploads/2026/10/internvl3_5-30b-input-api-market.mp4",
      "prompt": "Please describe this video in detail",
      "num_beams": 1,
      "max_frames": 8,
      "temperature": 0.2,
      "thinking_mode": false,
      "max_new_tokens": 1024,
      "repetition_penalty": 1
    }
  }'
```

Response (201 Created):
```json
{
  "id": "c4wgtt6ghhrna0d1d8hs6tbeqg",
  "version": "92713bb56e2e4827477501fc512d80d7287bc34e4ccb7c306afd9bdaef8b7eeb",
  "status": "starting",
  "created_at": "2026-10-23T19:01:34.685Z"
}
```

#### Check Prediction Status
```bash
curl -X GET "https://prod.api.market/api/v1/magicapi/vision-language-multimodal-ai-lucataco-internvl3-5-30b/predictions/{id}" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY"
```

Response when the prediction has finished (200 OK):
```json
{
  "id": "c4wgtt6ghhrna0d1d8hs6tbeqg",
  "status": "succeeded",
  "output": "In a sunlit park with lush greenery and people seated on benches, a monkey is seen riding a skateboard. The skateboard has a vibrant design with red wheels and a black deck. The monkey, with light brown fur, appears focused and balanced as it propels itself forward. The environment is lively yet relaxed, with dappled sunlight filtering through the trees, casting soft shadows on the pavement. The park is populated with visitors in casual attire, enjoying the day. The light suggests it's either morning or late afternoon. As the monkey skates, it maintains a steady pace, its gaze fixed ahead. The scene exudes a sense of fun and unexpectedness, highlighting the monkey's skill in an activity typically seen as human. The colors are warm and inviting, adding to the cheerful atmosphere of the clip.",
  "error": null,
  "created_at": "2026-10-23T19:01:34.685Z",
  "started_at": "2026-10-23T19:01:35.685Z",
  "completed_at": "2026-10-23T19:02:04.685Z"
}
```

`output` is a single text string with the model's answer.

### Status Codes
- **201 Created**: prediction created (POST `/predictions`). This is the normal success response.
- **200 OK**: prediction details returned (GET `/predictions/{id}`).
- **400 Bad Request**: invalid input, for example a `max_frames` above 32 or a `num_beams` above 5.
- **401 Unauthorized**: missing or invalid API key.
- **404 Not Found**: no prediction with that ID.
- **429 Too Many Requests**: rate limit exceeded.
- **500 Internal Server Error** or **503 Service Unavailable**: temporary server problem; retry later.

## Rate Limits
All plans allow 10 requests per second. Requests above that return `429 Too Many Requests`; back off before retrying and honour any `Retry-After` header. Polling a prediction with GET is free but still counts toward the per-second limit, so poll every 1 to 5 seconds rather than in a tight loop.

## Error Handling
- Check the HTTP status code of every request and read the response body for the error message.
- A prediction can also end with `status` set to `failed`; the `error` field then explains why. Failed predictions are not billed.
- Retry 429, 500 and 503 responses with exponential backoff. Fix the input before retrying a 400.

## Implementation Guide
### Example Code

#### JavaScript
```javascript
const apiKey = "YOUR_API_MARKET_KEY";
const baseUrl = "https://prod.api.market/api/v1/magicapi/vision-language-multimodal-ai-lucataco-internvl3-5-30b";

async function createAndWaitForPrediction(input) {
  const response = await fetch(`${baseUrl}/predictions`, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-api-market-key": apiKey
    },
    body: JSON.stringify({ version: "92713bb56e2e4827477501fc512d80d7287bc34e4ccb7c306afd9bdaef8b7eeb", input })
  });

  if (response.status !== 201) {
    throw new Error(`API error: ${response.status} ${await response.text()}`);
  }

  const prediction = await response.json();
  return pollForResults(prediction.id);
}

const delay = (ms) => new Promise((resolve) => setTimeout(resolve, ms));

async function pollForResults(predictionId) {
  const maxTimeout = 600000;
  const pollInterval = 1000;
  const startTime = Date.now();

  while (Date.now() - startTime < maxTimeout) {
    const response = await fetch(`${baseUrl}/predictions/${predictionId}`, {
      headers: { "x-api-market-key": apiKey }
    });

    if (!response.ok) {
      throw new Error(`API error: ${response.status} ${await response.text()}`);
    }

    const prediction = await response.json();

    if (prediction.status === "succeeded") {
      return prediction.output;
    }

    if (prediction.status === "failed" || prediction.status === "canceled") {
      throw new Error(`Prediction ${prediction.status}: ${prediction.error || "Unknown error"}`);
    }

    await delay(pollInterval);
  }

  throw new Error("Prediction timed out after 10 minutes");
}

createAndWaitForPrediction({
  "top_k": 50,
  "top_p": 0.95,
  "video": "https://blog.api.market/wp-content/uploads/2026/10/internvl3_5-30b-input-api-market.mp4",
  "prompt": "Please describe this video in detail",
  "num_beams": 1,
  "max_frames": 8,
  "temperature": 0.2,
  "thinking_mode": false,
  "max_new_tokens": 1024,
  "repetition_penalty": 1
}).then((text) => console.log(text));
```

#### Python
```python
import time
import requests

API_KEY = "YOUR_API_MARKET_KEY"
BASE_URL = "https://prod.api.market/api/v1/magicapi/vision-language-multimodal-ai-lucataco-internvl3-5-30b"


def create_and_wait_for_prediction(input_data):
    headers = {"Content-Type": "application/json", "x-api-market-key": API_KEY}
    response = requests.post(
        f"{BASE_URL}/predictions",
        headers=headers,
        json={"version": "92713bb56e2e4827477501fc512d80d7287bc34e4ccb7c306afd9bdaef8b7eeb", "input": input_data},
    )

    if response.status_code != 201:
        raise Exception(f"API error: {response.status_code} {response.text}")

    return poll_for_results(response.json()["id"])


def poll_for_results(prediction_id):
    max_timeout = 600
    poll_interval = 1
    headers = {"x-api-market-key": API_KEY}
    start_time = time.time()

    while time.time() - start_time < max_timeout:
        response = requests.get(f"{BASE_URL}/predictions/{prediction_id}", headers=headers)

        if response.status_code != 200:
            raise Exception(f"API error: {response.status_code} {response.text}")

        prediction = response.json()

        if prediction["status"] == "succeeded":
            return prediction["output"]

        if prediction["status"] in ("failed", "canceled"):
            raise Exception(f"Prediction {prediction['status']}: {prediction.get('error') or 'Unknown error'}")

        time.sleep(poll_interval)

    raise Exception("Prediction timed out after 10 minutes")


print(create_and_wait_for_prediction({
    "top_k": 50,
    "top_p": 0.95,
    "video": "https://blog.api.market/wp-content/uploads/2026/10/internvl3_5-30b-input-api-market.mp4",
    "prompt": "Please describe this video in detail",
    "num_beams": 1,
    "max_frames": 8,
    "temperature": 0.2,
    "thinking_mode": False,
    "max_new_tokens": 1024,
    "repetition_penalty": 1
}))
```

## FAQ
- **Should I send `images` or `video`?**
  - Send one of them per request. `images` takes one image URL; `video` takes a video URL, from which `max_frames` evenly spaced frames are used.
- **How do I get more detail from a long video?**
  - Raise `max_frames` (up to 32) so more of the clip is seen, and raise `max_new_tokens` for a longer answer. Both increase processing time.
- **What does `thinking_mode` do?**
  - It makes the model reason step by step before answering. This helps on harder questions but takes longer and can add units.
- **What format is the answer in?**
  - `output` is one plain text string.
- **What if my prediction fails?**
  - Failed predictions are not billed. Read the `error` field, adjust the input and try again.

## Contact Us
If you have any questions, reach out to us through the chat on the bottom right of the page. You can also email our support team at hello@magicapi.com.


Describe and question images and videos with InternVL3.5 30B, an open-source multimodal LLM with up to 32 video frames and an optional thinking mode.