Name: DeepSeek V3.1 API: Hybrid Reasoning LLM
Brand: magicapi
SKU: api-magicapi-hybrid-reasoning-llm-deepseek-ai-deepseek-v3-1
Availability: InStock
Author: API Market

Shashank Agarwal

# 💬 Hybrid Reasoning LLM API (DeepSeek V3.1)

## Overview
DeepSeek V3.1 is a large language model from DeepSeek. Its developer describes it as a hybrid thinking model: it can answer directly, or reason before it answers.

Send a text `prompt` and the model returns generated text. Set `thinking` to `medium` for enhanced reasoning, or leave it at `None` for the default behavior. `max_tokens` caps the length of the answer, and `temperature`, `top_p`, `presence_penalty` and `frequency_penalty` control sampling.

## Features
- **Hybrid reasoning**: `thinking` accepts `medium` (enhanced reasoning) or `None` (default behavior, no extra reasoning).
- **Output length cap**: `max_tokens` (default 1024) sets the maximum number of tokens the model generates.
- **Sampling controls**: `temperature` (default 0.1) and `top_p` (default 1).
- **Repetition controls**: `presence_penalty` and `frequency_penalty` (both default 0).
- **Text in, text out**: one `prompt` string per request; the answer comes back as text chunks you join.

## Use Cases
1. **Question answering**: explanations, definitions and summaries in plain text.
2. **Writing and rewriting**: drafts, emails, product copy and edits of existing text.
3. **Multi-step problems**: set `thinking` to `medium` for math, logic and planning questions.
4. **Coding questions**: ask for code snippets or explanations of code pasted into the prompt.

## Visual Examples
### Example Input
Prompt: "Explain the difference between a process and a thread in three short paragraphs, then give one example where each is the better choice."

Settings: `max_tokens` 1024, `temperature` 0.1, `top_p` 1, `presence_penalty` 0, `frequency_penalty` 0. `thinking` was left at its default (`None`).

### Example Output
The chunks of this run joined into one answer:

````markdown
Of course. Here is the explanation and examples.

A **process** is an independent, self-contained execution environment with its own memory space, system resources, and state. The operating system isolates processes from one another, meaning a crash in one process generally does not affect others. This isolation provides stability and security but comes with higher overhead for creation, termination, and communication (which requires inter-process communication or IPC mechanisms).

A **thread**, often called a "lightweight process," exists within a process and shares the process's memory space and resources. All threads of a single process can directly access shared data and communicate with each other efficiently. However, this lack of isolation means an error in one thread can corrupt data and crash the entire process. Threads are much faster to create and manage than processes due to their lower overhead.

The core difference is one of isolation versus sharing. A process provides a robust, isolated environment, while a thread enables efficient, lightweight concurrent execution and data sharing within a single application environment.

---

**Example where a process is better:** A **web browser**. Each browser tab is best run as a separate process. This provides critical security and stability; if a website in one tab crashes or has a malicious script, it is contained within its own process's memory sandbox and will not affect the other tabs or the browser's core UI.

**Example where a thread is better:** A **word processor**. One thread can handle user typing and UI updates, while another thread runs the spell and grammar checker in the background on the same document data. Using threads allows for a responsive interface and immediate sharing of the document's data in memory without the heavy overhead of process-level IPC.
````

## Input Parameters
| Field | Type | Required | Default | Allowed values | Description |
| --- | --- | --- | --- | --- | --- |
| `thinking` | string | No | `None` | `medium`, `None` | Reasoning effort level. Use `medium` for enhanced reasoning, or leave as `None` for the default behavior. |
| `prompt` | string | No | `Why are you better than Deepseek v3?` |  | Text prompt to send to the model |
| `max_tokens` | integer | No | `1024` |  | The maximum number of tokens the model should generate as output |
| `temperature` | number | No | `0.1` |  | The value used to modulate the next token probabilities |
| `top_p` | number | No | `1` |  | Top-p (nucleus) sampling |
| `presence_penalty` | number | No | `0` |  | Presence penalty |
| `frequency_penalty` | number | No | `0` |  | Frequency penalty |

## Pricing
- **Pay As You Go (PAYG) pricing**: you only pay for what you use.
- **Price**: $0.0026 per API unit.
- **1 API unit = 1,000 weighted tokens**; an output token counts 1, an input token counts 0.3334.
- That is **$0.86684 per million input tokens** and **$2.6 per million output tokens**.
- Each request is rounded up to a whole API unit, so the smallest charge per request is $0.0026.
- **Example cost**: the request above used 29 input tokens and 340 output tokens. 29 x 0.3334 + 340 = 349.67 weighted tokens, rounded up to 1 API unit = **$0.0026**.
- **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/hybrid-reasoning-llm-deepseek-ai-deepseek-v3-1`

### 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/hybrid-reasoning-llm-deepseek-ai-deepseek-v3-1/predictions" \
  -H "Content-Type: application/json" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY" \
  -d '{
    "version": "deepseek-ai/deepseek-v3.1",
    "input": {
      "top_p": 1,
      "prompt": "Explain the difference between a process and a thread in three short paragraphs, then give one example where each is the better choice.",
      "max_tokens": 1024,
      "temperature": 0.1,
      "presence_penalty": 0,
      "frequency_penalty": 0
    }
  }'
```

Response (201 Created):
```json
{
  "id": "pqd2a1mr9xrmy0d1c71r66qq2c",
  "version": "deepseek-ai/deepseek-v3.1",
  "status": "starting",
  "created_at": "2026-10-22T03:59:31.061Z"
}
```

#### Check Prediction Status
```bash
curl -X GET "https://prod.api.market/api/v1/magicapi/hybrid-reasoning-llm-deepseek-ai-deepseek-v3-1/predictions/{id}" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY"
```

Response when the prediction has finished (200 OK):
```json
{
  "id": "pqd2a1mr9xrmy0d1c71r66qq2c",
  "status": "succeeded",
  "output": [
    "Of course. Here is the explanation and examples.\n\nA **process** is an independent, self-contained execution environment with its own memory space, system resources, and state. The operating system isolates processes from one another, meaning a crash in one process generally does not affect others. This isolation provides stability and security but comes with higher overhead for creation, termination, and communication (which requires inter-process communication or IPC mechanisms).\n\nA **thread**, often called a \"lightweight process,\" exists within a process and shares the process's memory space and resources. All threads of a single process can directly access shared data and communicate with each other efficiently. However, this lack of isolation means an error in one thread can corrupt data and crash the entire process. Threads are much faster to create and manage than processes due to their lower overhead.\n\nThe core difference is one of isolation versus sharing. A process provides a robust, isolated environment, while a thread enables efficient, lightweight concurrent execution and data sharing within a single application environment.\n\n---\n\n**Example where a process is better:** A **web browser**. Each browser tab is best run as a separate process. This provides critical security and stability; if a website in one tab crashes or has a malicious script, it is contained within its own process's memory sandbox and will not affect the other tabs or the browser's core UI.\n\n**Example where a thread is better:** A **word processor**. One thread can handle user typing and UI updates, while another thread runs the spell and grammar checker in the background on the same document data. Using threads allows for a responsive interface and immediate sharing of the document's data in memory without the heavy overhead of process-level IPC."
  ],
  "error": null,
  "created_at": "2026-10-22T03:59:31.061Z",
  "started_at": "2026-10-22T03:59:31.073Z",
  "completed_at": "2026-10-22T03:59:38.053Z"
}
```

`output` is an array of text chunks. Join them in order to get the full answer. The example above shows the chunks from this run already joined into one string.

### 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 `thinking` value other than `medium` or `None`.
- **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/hybrid-reasoning-llm-deepseek-ai-deepseek-v3-1";

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: "deepseek-ai/deepseek-v3.1", 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.join("");
    }

    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_p": 1,
    "prompt": "Explain the difference between a process and a thread in three short paragraphs, then give one example where each is the better choice.",
    "max_tokens": 1024,
    "temperature": 0.1,
    "presence_penalty": 0,
    "frequency_penalty": 0
  }).then((output) => console.log(output));
```

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

API_KEY = "YOUR_API_MARKET_KEY"
BASE_URL = "https://prod.api.market/api/v1/magicapi/hybrid-reasoning-llm-deepseek-ai-deepseek-v3-1"

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": "deepseek-ai/deepseek-v3.1", "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 "".join(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_p": 1,
    "prompt": "Explain the difference between a process and a thread in three short paragraphs, then give one example where each is the better choice.",
    "max_tokens": 1024,
    "temperature": 0.1,
    "presence_penalty": 0,
    "frequency_penalty": 0
}))
```

## FAQ
- **How do I turn on reasoning?**
  - Set `thinking` to `medium`. The default, `None`, answers without the extra reasoning step.
- **Why does `output` come back as a list?**
  - The answer is returned as an array of text chunks. Join them in order (`output.join("")` in JavaScript, `"".join(output)` in Python).
- **Can I set a system prompt?**
  - There is no separate system prompt field. Put your instructions at the start of `prompt`.
- **How is a request billed?**
  - By weighted tokens: input tokens count 0.3334 and output tokens count 1, at $0.0026 per 1,000 weighted tokens, rounded up to a whole unit per request. Lower `max_tokens` to cap the output cost.
- **Is the 201 response from POST an error?**
  - No. 201 Created means the prediction was accepted. Use the returned `id` to poll GET `/predictions/{id}` until `status` is `succeeded`, `failed` or `canceled`.
- **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.


Generate text with DeepSeek V3.1, a hybrid model that answers directly or reasons first. Pay as you go, no monthly fee.