1. Overview: Two Calling Modes

On the Qwen3.7-Max pricing page, the Qianwen AI platform explicitly distinguishes two billing modes: Batch Chat and Batch File. These are not simply different price tiers but represent two fundamentally different API calling architectures.

2. Pricing Comparison

The Qwen3.7-Max pricing page lists complete price information for both modes. Note that the platform is currently in a limited-time 50% off promotion, making Batch Chat’s promotional price temporarily equal to Batch File’s regular price — but this is a temporary coincidence.

3. Batch Chat Explained

Batch Chat refers to synchronous real-time conversation API calls through the standard /v1/chat/completions endpoint. You send a message (or a set of messages), and the model generates and returns a response immediately. This is the most common API calling method.

3.1 Key Characteristics

3.2 Supported Advanced Features

As the standard conversation API, Batch Chat supports all advanced features of Qwen3.7-Max:

4. Batch File Explained

Batch File is the asynchronous batch inference service provided by the Qianwen AI platform. Users package multiple requests into a JSONL file and upload it. The system processes them in a background queue and generates results within 24 hours. The cost is only 50% of real-time calling.^[1]

4.1 How It Works

4.2 Input File Format

The input file is in JSONL format, with one independent request per line:

{"custom_id":"req-1","method":"POST","url":"/v1/chat/completions","body":{"model":"qwen3.7-max","messages":[{"role":"user","content":"Summarize quantum computing in two sentences."}]}}
{"custom_id":"req-2","method":"POST","url":"/v1/chat/completions","body":{"model":"qwen3.7-max","messages":[{"role":"user","content":"What is 2+2?"}]}}

4.3 Task Status Flow

4.4 Result File Format

After task completion, the system generates an output file (successful responses) and an error file (failure details, if any). Each line in the output file corresponds to the original request via custom_id:

{"id":"batch_req_xxx","custom_id":"req-1","response":{"status_code":200,"body":{"choices":[{"message":{"content":"..."}}],"usage":{...}}}}

5. Key Differences

6. Calling Methods & Code Examples

6.1 Batch Chat Calling Method

Batch Chat uses the standard OpenAI SDK compatible interface, directly calling chat.completions.create():

from openai import OpenAI
import os

client = OpenAI(
    api_key=os.getenv("DASHSCOPE_API_KEY"),
    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
)

# Batch Chat: synchronous real-time call
response = client.chat.completions.create(
    model="qwen3.7-max",
    messages=[{"role": "user", "content": "Summarize quantum computing in two sentences."}],
)

print(response.choices[0].message.content)

Supports streaming output and thinking mode:

# Streaming + Thinking mode
response = client.chat.completions.create(
    model="qwen3.7-max",
    messages=[{"role": "user", "content": "Prove that sqrt(2) is irrational"}],
    extra_body={"enable_thinking": True},
    stream=True,
)

for chunk in response:
    delta = chunk.choices[0].delta
    if hasattr(delta, "reasoning_content") and delta.reasoning_content:
        print(delta.reasoning_content, end="", flush=True)
    if hasattr(delta, "content") and delta.content:
        print(delta.content, end="", flush=True)

6.2 Batch File Calling Method

Batch File calling involves four steps: Upload File → Create Task → Query Status → Download Results.

1. Step 1: Upload JSONL File file_object = client.files.create( file=Path(“input.jsonl”), purpose=”batch” ) print(file_object.id) # file-batch-xxx The returned file ID can be reused. No need to re-upload identical files.
2. Step 2: Create Batch Task batch = client.batches.create( input_file_id=”file-batch-xxx”, endpoint=”/v1/chat/completions”, completion_window=”24h”, # 24h to 336h (14 days) metadata={ “ds_name”: “My batch job”, “ds_description”: “Batch processing task”, } ) print(batch.id) # batch_xxx endpoint must match the url in the input file. completion_window can be 24h to 336h (14 days).
3. Step 3: Query Task Status batch = client.batches.retrieve(“batch_xxx”) print(batch.status) # validating → in_progress → finalizing → completed Recommended polling interval: every 1-2 minutes. Download results when status becomes completed.
4. Step 4: Download Results content = client.files.content(“file-batch_output-xxx”) content.write_to_file(“result.jsonl”) Also download error_file_id for failed request details. Results are matched to original requests via custom_id.

Complete code example:

from openai import OpenAI
from pathlib import Path
import os, time

client = OpenAI(
    api_key=os.getenv("DASHSCOPE_API_KEY"),
    base_url="https://dashscope.aliyuncs.com/compatible-mode/v1",
)

# Step 1: Upload file
file_object = client.files.create(
    file=Path("input.jsonl"),
    purpose="batch"
)
print(f"File ID: {file_object.id}")

# Step 2: Create batch task
batch = client.batches.create(
    input_file_id=file_object.id,
    endpoint="/v1/chat/completions",
    completion_window="24h",
    metadata={"ds_name": "Qwen3.7-Max batch job"}
)
print(f"Task ID: {batch.id}")

# Step 3: Poll status
while True:
    batch = client.batches.retrieve(batch.id)
    print(f"Status: {batch.status}")
    if batch.status in ("completed", "failed", "expired", "cancelled"):
        break
    time.sleep(60)

# Step 4: Download results
if batch.status == "completed":
    content = client.files.content(batch.output_file_id)
    content.write_to_file("result.jsonl")
    print("Results saved to result.jsonl")

6.3 Preparing JSONL Input File

Use a CSV-to-JSONL script to batch-generate input files:

import csv, json

def build_messages(content):
    return [{"role": "user", "content": content}]

with open("input.csv") as fin, open("input.jsonl", "w") as fout:
    for row in csv.reader(fin):
        request = {
            "custom_id": row[0],
            "method": "POST",
            "url": "/v1/chat/completions",
            "body": {
                "model": "qwen3.7-max",
                "messages": build_messages(row[1])
            }
        }
        fout.write(json.dumps(request, ensure_ascii=False) + "\n")

6.4 Using the Console

In addition to API calls, Batch File can be operated through the Qianwen AI platform console:

1. Open the Batch API page Log in to the Qianwen AI platform console and navigate to the “Batch API” page.
2. Create a batch inference task Fill in task name and description, set max wait time (1-14 days), upload the JSONL input file.
3. Monitor and manage tasks View progress (processed/total requests) and status in the task list. Tasks in validating or in_progress status can be cancelled.
4. Download results After status becomes completed, download output and error files from the task detail page.

7. Use Case Recommendations

8. Important Notes