Qianfan LLM Workshop | Baidu AI Cloud International

Model list

Supported model services.

Click Display price to open list prices directly below the selected model row.

Service Name / ID	Service Rate Limit	Display Price	Actions
GLM-5.2svcp-07c4b1fd3190	RPM60TPM250000	Input$ 1.4 /M tokens Output$ 4.4 /M tokens Cache$ 0.26 /M tokens Discount70%-80%	Test
Kimi-K2.6svcp-73e5080f3170	RPM60TPM250000	Input$ 0.95 /M tokens Output$ 4 /M tokens Cache$ 0.16 /M tokens Discount80%-90%	Test
GLM-5.1svcp-99ef04eb3159	RPM60TPM150000	Input$ 1.4 /M tokens Output$ 4.4 /M tokens Cache$ 0.26 /M tokens Discount70%-80%	Test
DeepSeek-V4-Flashsvcp-2227d7533156	RPM60TPM150000	Input$ 0.14 /M tokens Output$ 0.28 /M tokens Cache$ 0.028 /M tokens Discount80%-90%	Test
DeepSeek-V4-Prosvcp-7221e5c83133	RPM60TPM150000	Input$ 1.69 /M tokens Output$ 3.38 /M tokens Cache$ 0.14 /M tokens Discount80%-90%	Test
CoBuddysvcp-54fda78a3131	RPM60TPM150000	Input$ 0.28 /M tokens Output$ 1.13 /M tokens Cache$ 0.07 /M tokens Discount-	Test
Qianfan-OCR-Fastsvcp-6fec24ea3065	RPM60TPM60000	No display price configured yet.	Test
Minimax-M2.5svcp-1bd4b29a3063	RPM60TPM250000	No display price configured yet.	Test
Kimi-K2.5svcp-b922ede93027	RPM60TPM250000	No display price configured yet.	Test
DeepSeek-V3.2svcp-0cdf61033026	RPM5000TPM1000000	Input$0.28/M tokens Output$0.42/M tokens Cache$0.028/M tokens Discount60%-70%	Test
GLM-5svcp-2bd235c33025	RPM60TPM250000	Input$1.0/M tokens Output$3.2/M tokens Cache$0.2/M tokens Discount70%-80%	Test
ERNIE-5.0svcp-e5522f7a3024	RPM60TPM150000	Input$1.4/M tokens Output$5.6/M tokens Cache- Discount-	Test

Access docs

Quick Start

Updated at: 2026-04-01 Document center / Qianfan Baidu / Get started / Quick Start

Welcome to the Baidu AI Cloud Qianfan AI Application Developer Center. This guide helps developers start building intelligent applications with text generation, conversational interaction, content analysis, and other model capabilities.

01

Create an API Key

Open API Key Console

Log in to Console - Security Authentication - API Key, then click Create API Key. The created API Key can be used for Qianfan AppBuilder, ModelBuilder, and AI Open Capabilities.

A single API Key supports calls to all three capabilities. Add permissions on the API Key page and select the service scopes you plan to call.

Request Header Authorization: Bearer bce-v3/ALTAK-xxxxx

02 Send a Large Model Chat Request

Once you have an API Key, send your first request with a REST client, the official Qianfan SDK, or an OpenAI-compatible SDK. Refer to the Model List to select the model parameter.

HTTP Request REST

curl --location 'https://api.baiduqianfan.ai/v1/chat/completions' \
--header 'Content-Type: application/json' \
--header 'Authorization: Bearer bce-v3/ALTAK-*********/614fb**********' \
--data '{
  "model": "ernie-3.5-8k",
  "messages": [
    {
      "role": "system",
      "content": "Platform Assistant"
    },
    {
      "role": "user",
      "content": "Hello"
    }
  ]
}'

SDK Request Python

from openai import OpenAI

client = OpenAI(
    api_key="bce-v3/ALTAK-KZke********/f1d6ee*************",
    base_url="https://api.baiduqianfan.ai/v1",
    default_headers={"appid": "app-xxxxxx"}
)

completion = client.chat.completions.create(
    model="ernie-4.0-turbo-8k",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"}
    ]
)

print(completion.choices[0].message)

Test console

API key connectivity and chat test.

The API Key is used only for this browser request and is not stored by this page.

API Key Model Message

Ready.

Methods

Chat Completions API

Method POST

URL https://api.baiduqianfan.ai/v1/chat/completions

Authentication Authorization: Bearer <API Key>

Parameter	Type	Required	Description
model	string	Yes	Model ID. For preset services, use the model value shown in the service details.
messages	array of objects	Yes	Chat context information. A single member indicates a single-turn conversation.
stream	boolean	No	Whether to return chunks through server-sent events.
temperature	number	No	Higher values make output more random; lower values make it more deterministic.
top_p	number	No	Controls output diversity.
max_tokens	integer	No	Maximum output tokens for the model.
tools	array of objects	No	Function descriptions that the model can trigger.
response_format	object	No	Response format options: text, json_object, or json_schema.
thinking	object	No	Deep thinking mode configuration.