Create a deployment-Alibaba Cloud Model Studio(Model Studio)-阿里云帮助中心

Create a model deployment task.

The model deployment API is currently available only in the China (Beijing) region. If you use another region, deploy models in the Model Studio console of that region.

Prerequisites

Read Introduction to model deployment and Deploy a model by using the API to understand how model deployment works and the basic workflow on Alibaba Cloud Model Studio.
Configure your API key for Model Studio. For details, see Get an API key.

Model deployment

Endpoint

POST https://dashscope.aliyuncs.com/api/v1/deployments

Request examples

Billing by provisioned throughput units (PTU)

Note

After running the deployment command below, billing starts immediately upon successful deployment—even if you have not yet called the model. Confirm the billing rules before deploying.

The PTU billing mode charges based on the duration of provisioned throughput usage. It suits scenarios requiring stable throughput guarantees, high concurrency, low latency, and predictable traffic. In this mode, throughput/concurrency and generation speed are preset by the platform and cannot be adjusted.

curl "https://dashscope.aliyuncs.com/api/v1/deployments" \
--header "Authorization: Bearer $DASHSCOPE_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
    "name": "my_qwen_flash",
    "model_name": "qwen-flash-2025-07-28",
    "plan": "ptu",
    "ptu_capacity": {
        "input_tpm": 10000,
	"output_tpm": 1000
    }
}'

Billing by model unit usage duration

Note

After running the deployment command below, billing starts immediately upon successful deployment—even if you have not yet called the model. Confirm the billing rules before deploying.
Model unit pay-as-you-go computing resources are allocated on a first-come, first-served basis. If purchase fails, you receive a full refund.

Select the model unit billing method. This mode charges based on model unit usage duration and suits large-scale inference workloads after model fine-tuning. Resources are dedicated, and performance and cost are flexible. Throughput/concurrency and generation speed are customizable.

curl "https://dashscope.aliyuncs.com/api/v1/deployments" \
--header "Authorization: Bearer $DASHSCOPE_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
    "name": "my_qwen_plus",
    "model_name": "qwen-plus-2025-12-01",
    "plan": "mu",
    "deploy_spec": "MU1",
    "enable_thinking": true,
    "capacity": 4,
    "max_context_length": 10000,
    "rpm_limit": 500,
    "tpm_limit": 1000
}'

The model unit deployment mode supports additional settings:

Configuration item	Configuration details
Service name	Custom name for the deployment service.
Select model	Select the model to deploy, including platform built-in models and fine-tuned models.
Model unit type	Select the deployment specification. Different specifications correspond to different compute power and performance.
Deployment replica count	Set the initial number of deployment replicas, which affects the concurrency capacity of the service.
Deployment template	Select a deployment template (such as "single-node deployment"). Different templates correspond to different resource configuration schemes. Available only under the model unit billing mode.
Configure model inference mode	For some models deployed in Model Unit mode, you can configure the inference mode, maximum context length, and more. Instruct - After deployment, the model performs inference in non-thinking mode. Thinking - After deployment, the model performs inference in thinking mode.
Maximum context length	The Model Unit deployment mode of some models supports this setting. The maximum context length depends on the model type.
Service rate limiting	The Model Unit deployment mode of some models supports this setting. You can limit the RPM and TPM of model invocations.

For details on setting these options via API, see Create a model deployment task using the API.

Billing by token usage

Select the token-based billing method. This mode charges based on token usage and suits cost-sensitive scenarios with low requirements for concurrency and latency. It offers the highest price advantage. Throughput/concurrency and generation speed are preset by the platform and cannot be adjusted.

curl "https://dashscope.aliyuncs.com/api/v1/deployments" \
--header "Authorization: Bearer $DASHSCOPE_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
    "model_name": "qwen3-8b-ft-202511132025-0260",
    "plan": "lora",
    "capacity": 1,
    "name": "qwen3-8b-ft"
}'

The capacity parameter has no effect but must be included. To scale up or down, go to the Model Studio model deployment console and submit a form request.

Request parameters

Parameter

Type

Location

Required

Description

model_name

String

body

Yes

The name of the model to deploy. This corresponds to the model ID in My Models . You can also get this ID from the output of the Create Training Job or Create Import Job operations.

plan

String

body

Yes

The deployment plan. The following billing methods are supported:

Billing method	Plan setting
Billing by model unit	`"plan": "mu"`
Billing by compute unit	`"plan": "cu"`
Provisioned throughput	`"plan": "ptu"`
LoRA shared deployment (billed by token usage)	`"plan": "lora"`

You can quickly find the supported deployment plans for a fine-tuned model in My Models .

Note

Fine-tuned CosyVoice models currently only support "plan": "mu".

name

String

body

Yes

The display name of the model in the console. If you do not specify this parameter, the value of model_name is used as the deployment name.

capacity

Integer

body

Required only when "plan": "mu" is specified. Specifies the number of resource units for the deployment. The value must be an integer multiple of base_capacity. The constraints vary based on the deploy_spec value. For example, for MU2, the value must be a multiple of 8, while for MU5, it can be 1. Example: "capacity": 1.

Note

CosyVoice models currently provide the following two deployment templates with corresponding capacity constraints:

single-node deployment: capacity must be an integer multiple of 1, such as 1, 2, 3, 4, or 5.
single-node deployment - flagship complex inference edition: capacity must be an integer multiple of 8, such as 8, 16, 24, or 32.

billing_method

String

body

Required only when "plan": "mu" is specified. Currently, only "POST_PAY" (Post-paid) is supported. Example: "billing_method": "POST_PAY".

deploy_spec

String

body

This setting is applicable only when "plan": "mu" is specified.

For details about feature support, see Feature support for model unit deployment.

This parameter is required when "plan": "mu" is specified. Example: "deploy_spec": "MU1".

Note

You can get this value from the template_id field returned by the Get Deployable Model List operation.

enable_thinking

Boolean

body

Supported by some models. You can set this to true or false.

max_context_length

Number

body

Supported by some models. Example: "max_context_length": 131072.

rpm_limit

Number

body

Supported by some models. Specifies the maximum number of requests per minute (RPM).

tpm_limit

Number

body

Supported by some models. Specifies the maximum number of tokens per minute (TPM).

ptu_capacity

Object

body

This setting is applicable only when "plan": "ptu" is specified.

For details about feature support, see Feature Support for PTU Deployment.

If you do not specify this parameter, the system defaults to 10,000 input_tpm and 1,000 output_tpm.

Example: "ptu_capacity": { "input_tpm": 10000, "output_tpm": 1000 }.

ptu_capacity.input_tpm

Number

body

Supported by all models. Specifies the maximum number of input tokens per minute (TPM).

ptu_capacity.output_tpm

Number

body

Supported by all models. Specifies the maximum number of output tokens per minute (TPM).

ptu_capacity.thinking_output_tpm

Number

body

Supported by some models. Specifies the maximum number of provisioned thinking output tokens per minute (TPM).

suffix

String

body

After a model is deployed, a new model name is generated. The suffix parameter specifies the suffix for this new name. It must be globally unique and have a maximum length of 8 characters. You can omit the suffix for the first deployment of a model. If you deploy the same model multiple times, you must specify a unique suffix for each deployment.

See the deployed_model output parameter for more information.

aigc_config

Object

body

The prompt configuration for a LoRA-based video generation model. This parameter applies only to Wan video generation models (the wan series). This parameter is not supported by wan2.7-image-pro. This object includes the following fields:

use_input_prompt (Boolean, required): Specifies whether to use the prompt provided by the caller. false: Use the preset template below. true: Use the prompt passed in through the inference API. The default value is false.
prompt (String, Required): The preset prompt template. This parameter takes effect only when use_input_prompt=false, overriding the prompt passed to the inference API.
lora_prompt_default (String, required): The fallback prompt template. Used if the prompt field is invalid or its configuration fails.

The aigc_config.prompt parameter

Why set this parameter?

If this parameter is not set, you must manually write a prompt for each model call. After you set it, you only need to upload an image when calling the fine-tuned model. The system then automatically provides the special effect instruction, so you do not have to enter a prompt manually.

How to set this parameter

prompt = task instruction (instructs the model on its task) + output template (defines the structure for the output) + example (provides a complete example for the model to follow).

Provide a video description based on the image content. The description must include "Then, start showing the s86b5p money rain special effect. Countless huge US dollar bills (beige with dark green patterns) pour down like a torrential rain, densely striking and swirling around her (him). The bills continue to fall as the camera slowly zooms in. She (he) spreads her (his) arms upward, neck slightly tilted back, with a surprised and joyful expression, completely immersed in this wild money rain."
Output template:
The video begins by showing [subject description]. [environment description]. Then, start showing the s86b5p money rain special effect. Countless huge US dollar bills (beige with dark green patterns) pour down like a torrential rain, densely striking and swirling around her (him). The bills continue to fall as the camera slowly zooms in. She (he) spreads her (his) arms upward, neck slightly tilted back, with a surprised and joyful expression, completely immersed in this wild money rain.
Example:
The video begins by showing a young woman on a beach. Her hair is wet, dark brown, curly, and slightly messy. She has a bright smile on her face. The background shows crashing waves and distant mountains. Then, start showing the s86b5p money rain special effect. Countless huge US dollar bills (beige with dark green patterns) pour down like a torrential rain, densely striking and swirling around her. The bills continue to fall as the camera slowly zooms in. She spreads her arms upward, neck slightly tilted back, with a surprised and joyful expression, completely immersed in this wild money rain.

For details about the output template, see the Wan Video Fine-tuning Guide - Output Templates.

First frame

Replace <YOUR_MODEL_NAME> with the value of the finetuned_output output parameter from your training job.

curl --location 'https://dashscope.aliyuncs.com/api/v1/deployments' \
--header "Authorization: Bearer $DASHSCOPE_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
    "model_name": "<YOUR_MODEL_NAME>",
    "aigc_config": {
        "use_input_prompt": false,
        "prompt": "Provide a video description based on the image content. The description must include \"Then the s86b5p money rain effect begins, where countless huge US dollar bills (beige with a dark green pattern) pour down like a torrential rain, falling heavily on and surrounding them. The bills continue to fall as the camera slowly zooms in. They stretch their arms upward, neck slightly tilted back, with a surprised expression, completely immersed in this wild money rain.\"\nOutput Template:\nThe video begins with a shot of [subject description]. [Environment description]. Then the s86b5p money rain effect begins...\nExample:\nThe video begins with a shot of a young woman on a beach...",
        "lora_prompt_default": "Then the s86b5p money rain effect begins, where countless huge US dollar bills (beige with a dark green pattern) pour down like a torrential rain, falling heavily on and surrounding the main character. The bills continue to fall as the camera slowly zooms in. The main character stretches their arms upward, neck slightly tilted back, with a surprised expression, completely immersed in this wild money rain."
    },
    "capacity": 1,
    "plan": "lora"
}'

First and last frames

curl --location 'https://dashscope.aliyuncs.com/api/v1/deployments' \
--header "Authorization: Bearer $DASHSCOPE_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
    "model_name": "<YOUR_MODEL_NAME>",
    "aigc_config": {
        "use_input_prompt": false,
        "prompt": "Provide a video description based on the image content. The description must include \"Then she begins the s86b5p transformation.\"\nOutput Template:\nThe video begins with a shot of [subject description]. [Environment description]. Then she begins the s86b5p transformation.\nExample:\nThe video begins with a shot of a young woman in an outdoor setting...",
        "lora_prompt_default": "Then she begins the s86b5p transformation."
    },
    "capacity": 1,
    "plan": "lora"
}'

Image generation

The wan2.7-image-pro model does not support the aigc_config parameter during deployment.

curl --location 'https://dashscope.aliyuncs.com/api/v1/deployments' \
--header "Authorization: Bearer $DASHSCOPE_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
    "model_name": "<YOUR_MODEL_NAME>",
    "capacity": 1,
    "plan": "lora"
}'

View supported features and billing.

Usage duration billing (Provisioned Throughput)

Cost = Usage duration × (Input TPM unit price × Input TPM + Output TPM unit price × Output TPM)

Postpaid is calculated by hour: the usage duration unit is hours, and the unit price takes the "1 hour continuous" column in the table below; prepaid is calculated by day: the usage duration unit is days, and the unit price takes the "1 day continuous" column in the table below.

Prepaid orders take effect immediately after payment, valid for N days until 23:59 on day N. If ordered after 22:00, the expiration date is automatically extended by 1 day.
After a prepaid order expires, the service will be stopped with a 2-hour delay, and resources will be retained for 14 hours after stopping before being released.
Prepaid orders cannot be terminated early.
For post-paid billing, if your account is in arrears, the deployed resources will be retained and billed for 24 hours, during which the service can still be used normally. After 24 hours, the system stops billing, the model deployment enters an arrears state, and the underlying resources will be deleted, but the model deployment task will be retained. After you pay off the arrears, the system will reallocate resources and resume usage (fees will continue to accrue after resumption). If you do not want to continue incurring charges, you can delete the model deployment task; once deleted successfully, billing will stop.

When the model input exceeds the maximum input Token, the relevant call will automatically switch to the pay-as-you-go mode of the current model; when the purchased TPM is exceeded, it is handled according to the overflow policy selected at creation ("Auto-overflow" switches to pay-as-you-go, "PTU capacity only" returns 429). At this time, inference performance may degrade and will be governed by the public traffic of the current snapshot model in the workspace, and the fee is charged according to the model invocation (pay-as-you-go) standard.

At this time (only under the "Auto-overflow" policy), the call API response Header will include: x-dashscope-ptu-overflow:true.
For TPM statistics, go to: Model Monitoring (Beijing).

For specific fee-reduction and refund rules for scale-in (downgrade) scenarios, please refer to: Refund rules for configuration downgrades.

Note

PTU deployment supports stepped capacity coefficients and cache discounts for long inputs. For details, see Provisioned Throughput long input and cache.

North China 2 (Beijing)

Qwen

Model name	Model code	Max input tokens	Postpaid input Per 10K TPM/hour	Postpaid output Per 1K TPM/hour	Prepaid input Per 10K TPM/day	Prepaid output Per 1K TPM/day
Qwen3.8-Max	qwen3.8-max	128K	¥28.8	¥8.64	¥345.6	¥103.68

Qwen3.7-Max-2026-05-20	qwen3.7-max-2026-05-20	256K	¥28.8	¥8.64	¥345.6	¥103.68
Qwen3.7-Plus-2026-05-26	qwen3.7-plus-2026-05-26	256K	¥4.8	¥1.92	¥57.6	¥23.04

Qwen3.6-Plus-2026-04-02	qwen3.6-plus-2026-04-02	128K	¥4.8	¥2.88	¥57.6	¥34.56

Qwen3.5-Plus-2026-04-20	qwen3.5-plus-2026-04-20	128K	¥1.92	¥1.15	¥23.04	¥13.82

Qwen3-Max-2025-09-23	qwen3-max-2025-09-23	128K	¥7.68	¥3.08	¥92.16	¥36.96

Qwen-Flash-2025-07-28	qwen-flash-2025-07-28	128K	¥0.36	¥0.36	¥4.32	¥4.32
Qwen-Plus-2025-12-01	qwen-plus-2025-12-01	128K	¥1.92	Non-thinking: ¥0.48 Thinking: ¥1.92	¥23.04	Non-thinking: ¥5.76 Thinking: ¥23.04

DeepSeek

Model name	Model code	Max input tokens	Postpaid input Per 10K TPM/hour	Postpaid output Per 1K TPM/hour	Prepaid input Per 10K TPM/day	Prepaid output Per 1K TPM/day
DeepSeek-v4-Flash	deepseek-v4-flash	256K	¥3.6	¥0.72	¥43.2	¥8.64
DeepSeek-v4-Pro	deepseek-v4-pro	256K	¥43.2	¥8.64	¥518.4	¥103.68
DeepSeek-v3.2	deepseek-v3.2	64K	¥7.2	¥1.08	¥86.4	¥12.96
DeepSeek-v3	deepseek-v3	64K	¥7.2	¥2.88	¥86.4	¥34.56

GLM

Model name

Model code

Max input tokens

Postpaid input

Per 10K TPM/hour

Postpaid output

Per 1K TPM/hour

Prepaid input

Per 10K TPM/day

Prepaid output

Per 1K TPM/day

GLM-5.2

glm-5.2

¥28.8

¥10.08

¥345.6

¥120.96

GLM-5.1

glm-5.1

64K

¥21.6

¥8.64

¥259.2

¥103.68

Qwen-VL

Model name

Model code

Max input tokens

Postpaid input

Per 10K TPM/hour

Postpaid output

Per 1K TPM/hour

Prepaid input

Per 10K TPM/day

Prepaid output

Per 1K TPM/day

Qwen3-VL-Plus-2025-09-23

qwen3-vl-plus-2025-09-23

128K

¥2.4

¥28.8

Singapore

Qwen

Model name	Model code	Max input tokens	Postpaid input Per 10K TPM/hour	Postpaid output Per 1K TPM/hour	Prepaid input Per 10K TPM/day	Prepaid output Per 1K TPM/day
Qwen3.8-Max	qwen3.8-max	128K	¥35.97	¥10.79	¥431.7	¥129.5

Qwen3.7-Max-2026-05-20	qwen3.7-max-2026-05-20	256K	¥44.97	¥13.49	¥539.6	¥161.87
Qwen3.7-Plus-2026-05-26	qwen3.7-plus-2026-05-26	256K	¥7.19	¥2.88	¥86.3	¥34.53

Qwen3.6-Plus-2026-04-02	qwen3.6-plus-2026-04-02	128K	¥9	¥5.4	¥107.9	¥64.75

Qwen3.5-Plus-2026-04-20	qwen3.5-plus-2026-04-20	128K	¥7.2	¥4.32	¥86.3	¥51.8

DeepSeek

Model name	Model code	Max input tokens	Postpaid input Per 10K TPM/hour	Postpaid output Per 1K TPM/hour	Prepaid input Per 10K TPM/day	Prepaid output Per 1K TPM/day
DeepSeek-v4-Flash	deepseek-v4-flash	256K	¥5.4	¥1.08	¥64.8	¥12.95
DeepSeek-v4-Pro	deepseek-v4-pro	256K	¥64.75	¥12.95	¥777	¥155.4
DeepSeek-v3.2	deepseek-v3.2	64K	¥15.4	¥4.61	¥184.5	¥55.36

GLM

Model name

Model code

Max input tokens

Postpaid input

Per 10K TPM/hour

Postpaid output

Per 1K TPM/hour

Prepaid input

Per 10K TPM/day

Prepaid output

Per 1K TPM/day

GLM-5.2

glm-5.2

¥37.8

¥11.87

¥453.3

¥142.45

GLM-5.1

glm-5.1

64K

¥37.77

¥11.87

¥453.3

¥142.45

Qwen-VL

Model name

Model code

Max input tokens

Postpaid input

Per 10K TPM/hour

Postpaid output

Per 1K TPM/hour

Prepaid input

Per 10K TPM/day

Prepaid output

Per 1K TPM/day

Qwen3-VL-Plus-2025-09-23

qwen3-vl-plus-2025-09-23

128K

¥3.6

¥2.88

¥43.2

¥34.53

Billing by usage duration (model unit)

Cost = Usage duration (hours) × Number of model units × Model unit price

"Model unit price" takes the "Hourly unit price" column in the table below for pay-as-you-go scenarios; for monthly prepaid billing, the formula becomes Number of months × Number of model units × Monthly unit price.

For the first month of a prepaid purchase, if you cancel early within the first month, the daily unit price (≈ Monthly unit price / 30) is billed at 1.2× (less than one day is billed as one day)

Note

Compute resources under the model unit pay-as-you-go method are first-come, first-served. If the purchase fails, a full refund is issued.

Text generation

Qwen

Model name	Model code	Model unit specification	Hourly unit price (CNY) Minimum billing: minute	Monthly unit price (CNY) Minimum billing: day
Qwen3.6-35B-A3B	qwen3.6-35b-a3b	MU1 x 8	¥432	¥208,944
		MU2 x 8	¥504	¥240,288
		MU3 x 8	¥1,096	¥527,752
		MU9 x 1	¥51	¥24,600
Qwen3.6-Flash-2026-04-16	qwen3.6-flash-2026-04-16	MU1 x 2	¥108	¥52,236
Qwen3.6-Flash-2026-04-16	qwen3.6-flash-2026-04-16	MU3 x 8	¥1,096	¥527,752
Qwen3.6-Plus-2026-04-02	qwen3.6-plus-2026-04-02	MU1 x 8 MU1 x 16 (PD separation mode)	¥432 PD separation mode: ¥864	¥208,944 PD separation mode: ¥417,888

Qwen3.5-397B-A17B	qwen3.5-397b-a17b	MU3 x 8 MU3 x 16 (PD separation mode)	¥1,096 PD separation mode: ¥2,192	¥527,752 PD separation mode: ¥1,055,504
Qwen3.5-397B-A17B	qwen3.5-397b-a17b	MU6 x 16	¥400	¥193,424
Qwen3.5-122B-A10B	qwen3.5-122b-a10b	MU1 x 4	¥216	¥104,472
Qwen3.5-122B-A10B	qwen3.5-122b-a10b	MU6 x 16	¥400	¥193,424
Qwen3.5-35B-A3B	qwen3.5-35b-a3b	MU1 x 2	¥108	¥52,236
		MU2 x 8	¥504	¥240,288
		MU3 x 8	¥1,096	¥527,752
		MU9 x 1	¥51	¥24,600
Qwen3.5-27B	qwen3.5-27b	MU2 x 8	¥504	¥240,288
		MU3 x 8	¥1,096	¥527,752
		MU8 x 1	¥47	¥22,400
		MU9 x 1	¥51	¥24,600
Qwen3.5-9B	qwen3.5-9b	MU2 x 2	¥126	¥60,072
Qwen3.5-Flash-2026-02-23	qwen3.5-flash-2026-02-23	MU1 x 2	¥108	¥52,236
Qwen3.5-Plus-2026-02-15	qwen3.5-plus-2026-02-15	MU1 x 8 MU1 x 16 (PD separation mode)	¥432 PD separation mode: ¥864	¥208,944 PD separation mode: ¥417,888
		MU2 x 8	¥504	¥240,288
		MU3 x 8 MU3 x 16 (PD separation mode)	¥1,096 PD separation mode: ¥2,192	¥527,752 PD separation mode: ¥1,055,504

Qwen3-235B-A22B-Instruct-2507	qwen3-235b-a22b-instruct-2507	MU1 x 4	¥216	¥104,472
		MU2 x 8	¥504	¥240,288
		MU3 x 8	¥1,096	¥527,752
Qwen3-32B	qwen3-32b	MU6 x 16	¥400	¥193,424
Qwen3-30B-A3B-Thinking-2507	qwen3-30b-a3b-thinking-2507	MU1 x 2	¥108	¥52,236
Qwen3-8B	qwen3-8b	MU1 x 2	¥108	¥52,236
Qwen3-8B	qwen3-8b	MU2 x 2	¥126	¥60,072
Qwen3-4B	qwen3-4b	MU1 x 2	¥108	¥52,236
Qwen3-4B	qwen3-4b	MU5 x 1	¥21	¥10,139
Qwen3-Embedding-0.6B	qwen3-embedding-0.6b	MU5 x 1	¥21	¥10,139
Qwen3-Embedding-0.6B	qwen3-embedding-0.6b	MU6 x 1	¥25	¥12,089
Qwen3-MoE-Rerank-0.6B	qwen3-moe-rerank-0.6b	MU5 x 1	¥21	¥10,139
Qwen3-Rerank-0.6B	qwen3-rerank-0.6b	MU5 x 1	¥21	¥10,139
Qwen3-Rerank-0.6B	qwen3-rerank-0.6b	MU6 x 1	¥25	¥12,089
Qwen3-Max-2025-09-23	qwen3-max-2025-09-23	MU2 x 8	¥504	¥240,288
Qwen3-Max-2025-09-23	qwen3-max-2025-09-23	MU3 x 8	¥1,096	¥527,752
Qwen3-Rerank	qwen3-rerank	MU5 x 1	¥21	¥10,139

Qwen2.5-72B (Open Source)	qwen2.5-72b-instruct	MU1 x 8	¥432	¥208,944
Qwen2.5-14B (Open Source)	qwen2.5-14b-instruct	MU1 x 2	¥108	¥52,236
Qwen2.5-7B (Open Source)	qwen2.5-7b-instruct	MU1 x 2	¥108	¥52,236
Qwen2.5-7B (Open Source)	qwen2.5-7b-instruct	MU5 x 1	¥21	¥10,139

Qwen-Plus-2025-07-28	qwen-plus-2025-07-28	MU1 x 4 MU1 x 16 (PD separation mode)	¥216 PD separation mode: ¥864	¥104,472 PD separation mode: ¥417,888
Qwen-Plus-2025-12-01	qwen-plus-2025-12-01	MU1 x 4	¥216	¥104,472
Qwen-Plus-Character-2025-11-06	qwen-plus-character-2025-11-06	MU1 x 4	¥216	¥104,472

GLM

Model name	Model code	Model unit specification	Hourly unit price (CNY) Minimum billing: minute	Monthly unit price (CNY) Minimum billing: day
GLM-5.1	glm-5.1	MU2 x 8	¥504	¥240,288
		MU3 x 16 (PD separation mode)	PD separation mode: ¥2,192	PD separation mode: ¥1,055,504
		MU6 x 16	¥400	¥193,424
GLM-5	glm-5	MU3 x 16 (PD separation mode)	PD separation mode: ¥2,192	PD separation mode: ¥1,055,504
GLM-4.7	glm-4.7	MU6 x 32 (PD separation mode)	PD separation mode: ¥800	PD separation mode: ¥386,848

DeepSeek

Model name	Model code	Model unit specification	Hourly unit price (CNY) Minimum billing: minute	Monthly unit price (CNY) Minimum billing: day
DeepSeek-v4-Flash	deepseek-v4-flash	MU1 x 8	¥432	¥208,944
DeepSeek-v4-Flash	deepseek-v4-flash	MU3 x 8	¥1,096	¥527,752
DeepSeek-v3.2	deepseek-v3.2	MU2 x 16 (PD separation mode)	PD separation mode: ¥1,008	PD separation mode: ¥480,576

More models

Model name

Model code

Model unit specification

Hourly unit price (CNY)

Minimum billing: minute

Monthly unit price (CNY)

Minimum billing: day

Kimi-K2.5

kimi-k2.5

MU2 x 8

¥504

¥240,288

Model type:

Instruct - After deployment, the model performs inference in non-thinking mode.
Thinking - After deployment, the model performs inference in thinking mode.

Model deployment type:

PD separation mode - Reduces first-token latency and increases throughput.

For models deployed in this mode, during inference the first-token computation (Prefill) and subsequent-token computation (Decode) — two computation phases — are split onto different compute nodes.

Multimodal

Qwen-VL

Model name	Model code	Model unit specification	Hourly unit price (CNY) Minimum billing: minute	Monthly unit price (CNY) Minimum billing: day
Qwen3-VL-235B-A22B-Thinking	qwen3-vl-235b-a22b-thinking	MU1 x 8	¥432	¥208,944
		MU2 x 8	¥504	¥240,288
		MU3 x 8	¥1,096	¥527,752
Qwen3-VL-32B-Instruct	qwen3-vl-32b-instruct	MU2 x 8	¥504	¥240,288
Qwen3-VL-32B-Instruct	qwen3-vl-32b-instruct	MU3 x 8	¥1,096	¥527,752
Qwen3-VL-8B-Instruct	qwen3-vl-8b-instruct	MU1 x 2	¥108	¥52,236
Qwen3-VL-8B-Instruct	qwen3-vl-8b-instruct	MU5 x 1	¥21	¥10,139
Qwen3-VL-4B-Instruct	qwen3-vl-4b-instruct	MU1 x 2	¥108	¥52,236
Qwen3-VL-2B-Instruct	qwen3-vl-2b-instruct	MU5 x 1	¥21	¥10,139
Qwen3-VL-Embedding-2B	qwen3-vl-embedding-2b	MU5 x 1	¥21	¥10,139
Qwen3-VL-Flash-2025-10-15	qwen3-vl-flash-2025-10-15	MU1 x 4	¥216	¥104,472
Qwen3-VL-Plus-2025-09-23	qwen3-vl-plus-2025-09-23	MU1 x 4	¥216	¥104,472

Qwen-VL-Max-2025-08-13	qwen-vl-max-2025-08-13	MU6 x 4	¥100	¥48,356

Qwen Omni

Model name

Model code

Model unit specification

Hourly unit price (CNY)

Minimum billing: minute

Monthly unit price (CNY)

Minimum billing: day

Model type:

Instruct - After model deployment, inference is performed in non-thinking mode.
Thinking - After model deployment, inference is performed in thinking mode.
Instruct/Thinking - You can choose whether to enable thinking mode during model deployment.

Speech synthesis

CosyVoice

Model name	Model code	Model unit specification	Hourly unit price (CNY)	Monthly unit price (CNY)
cosyvoice-v3-flash	cosyvoice-v3-flash	MU5	¥21	¥10,139

By model Token usage

Fee = Model input token count × Model input unit price + Model output token count × Model output unit price (minimum billing unit: 1 token)

Billing by model Token usage is supported only after you complete SFT efficient training on the following base models and obtain a custom model.

Qwen

Base model	Model code	Input CNY/1,000 tokens	Output CNY/1,000 tokens
Qwen3.5-27B In beta	qwen3.5-27b	¥0.0018	¥0.0048

Qwen3-32B	qwen3-32b	¥0.002	Non-thinking mode:¥0.008 Thinking mode:¥0.02
Qwen3-14B	qwen3-14b	¥0.001	Non-thinking mode:¥0.004 Thinking mode:¥0.01
Qwen3-8B	qwen3-8b	¥0.0005	Non-thinking mode:¥0.002 Thinking mode:¥0.005
Qwen3-4B-Instruct-2507	qwen3-4b-instruct-2507	¥0.0003	Non-thinking mode:¥0.0012 Thinking mode:¥0.003

Qwen2.5-Open-Source-72B	qwen2.5-72b-instruct	¥0.004	¥0.012
Qwen2.5-Open-Source-32B	qwen2.5-32b-instruct	¥0.002	¥0.006
Qwen2.5-Open-Source-14B	qwen2.5-14b-instruct	¥0.001	¥0.003
Qwen2.5-Open-Source-7B	qwen2.5-7b-instruct	¥0.0005	¥0.001

Qwen2-Open-Source-7B	qwen2-7b-instruct	¥0.001	¥0.002

Qwen-VL

Base model	Model code	Input CNY/1,000 tokens	Output CNY/1,000 tokens
Qwen3-VL-8B-Instruct	qwen3-vl-8b-instruct	¥0.0005	¥0.002

Qwen2.5-VL-72B	qwen2.5-vl-72b-instruct	¥0.016	¥0.048
Qwen2.5-VL-32B	qwen2.5-vl-32b-instruct	¥0.008	¥0.024
Qwen2.5-VL-7B	qwen2.5-vl-7b-instruct	¥0.002	¥0.005
Qwen2.5-VL-3B-Instruct	qwen2.5-vl-3b-instruct	¥0.0012	¥0.0036

Deployment troubleshooting and performance tuning

The max-num-seqs parameter is not configurable

For both Billing by provisioned throughput units (PTU) and Billing by model unit, Model Studio presets throughput, concurrency, and generation speed. You cannot directly tune vLLM engine parameters such as max-num-seqs, and this operation does not expose such a parameter. Under Billing by model unit, you can indirectly control throughput by choosing a Model unit type (the deploy_spec request parameter) and adjusting the Deployment replica count (the capacity request parameter).

Deployment templates and resource isolation

Deployment templates currently support single-node deployment only (Single-node deployment - Enhanced general-purpose inference). Multi-GPU instance isolation is not supported. If you need isolated resources, use Billing by model unit, which provides dedicated compute resources.

Handling throttling errors

When concurrent requests exceed the rate limit, the operation returns HTTP 429 with the error code Throttling.RateQuota and the message Requests rate limit exceeded, please try again later. To resolve this issue:

Billing by provisioned throughput units (PTU): adjust input_tpm and output_tpm in ptu_capacity (Input kTPM and Output kTPM in the console), or reduce the request rate. In this mode, the overflow policy can be Auto-overflow (switches to pay-as-you-go) or PTU capacity only, where requests beyond the provisioned capacity return 429 directly.
Billing by model unit: adjust rpm_limit and tpm_limit, or reduce the request rate.

Tuning the maximum context length

Under Billing by model unit, you can set the Maximum context length through the max_context_length request parameter, which accepts values from 1 to 262144. The actual upper limit depends on the deployed model. This parameter caps the context size of a single request, which limits per-request memory usage and reduces the risk of out-of-memory (OOM) errors. When you process large volumes of image tasks, set this parameter based on the image resolution and the number of images per request.

Response example

The command returns the following:

{
  "request_id": "f2ae64f7-83cc-410c-bc0b-840443f7eb86",
  "output": {
    "deployed_model": "emo-35b3f106-sample01",
    "gmt_create": "2025-06-17T11:00:38.68",
    "gmt_modified": "2025-06-17T11:00:38.68",
    "status": "PENDING",
    "model_name": "emo",
    "base_model": "emo",
    "base_capacity": 1,
    "capacity": 1,
    "ready_capacity": 0,
    "workspace_id": "llm-v71tlv3d***",
    "charge_type": "post_paid",
    "creator": "175805416***",
    "modifier": "175805416***"
  }
}

Response parameters

Parameter	Type	Description
request_id	String	The ID of the request.
output	Object	Details of the deployment task.
deployed_model	String	A unique identifier for the deployed model. This ID is used for API operations, such as querying deployment details, modifying deployment rate limiting, deployment scaling, and deleting deployments, and is also passed as an SDK parameter when you invoke the model.
gmt_create	String	The creation time of the deployment task.
gmt_modified	String	The last modification time of the deployment task.
status	String	The status of the deployment task. PENDING: The task is being created. UPDATING: The task is being updated. RUNNING: The deployment task is running, and the deployed model can process requests. STOPPED: The deployment task is stopped and is not billed. DELETING: The task is being deleted. FAILED: The creation or update of the task failed.
model_name	String	The name of the model used in the deployment task.
base_model	String	The ID of the base model used in the deployment task.
base_capacity	Number	The minimum number of resource units required to run the base model.
capacity	Number	The number of resource units used by the deployment task.
ready_capacity	Number	The number of resource units that are ready to process requests immediately. Resource initialization speed or hardware status can limit this value.
workspace_id	String	The ID of the deployment task's workspace.
charge_type	String	The billing method for the deployment task. `post_paid`: Post-paid.
creator	String	The UID of the user who created the deployment task.
modifier	String	The UID of the user who last modified the deployment task.
plan	String	The billing model for the deployment task. This parameter is not returned for some billing models.
Returned only for Model Unit deployments.
model_unit_spec	String	The model unit specification.
enable_thinking	Boolean	Specifies if Thinking mode is enabled. This feature is only available for certain models.
max_context_length	Number	The maximum context length.
rpm_limit	Number	The maximum number of requests per minute (RPM).
tpm_limit	Number	The maximum number of tokens per minute (TPM).
Returned only for provisioned throughput (PTU) deployments
ptu_capacity	Object	This parameter takes effect only when `"plan": "ptu"` is set. Example: `"ptu_capacity": { "input_tpm": 10000, "output_tpm": 1000 }`.
ptu_capacity.input_tpm	Number	The maximum number of input tokens per minute (TPM) for the deployed model. This feature is supported by all models.
ptu_capacity.output_tpm	Number	The maximum number of output tokens per minute (TPM) for the deployed model. This feature is supported by all models.
ptu_capacity.thinking_output_tpm	Number	The maximum number of thinking output tokens per minute (TPM) for the deployed model. This feature is only available for certain models.

Error response

Response example

{
    "request_id": "ca218d57-b91b-46b2-bd35-c41c6287bcf4",
    "message": "Model: qwen-plus-20230703-cx7f not found!",
    "code": "NotFound"
}

Response parameters

Parameter	Type	Description
request_id	String	The unique ID of the request.
code	String	The error code.
message	String	The error message.

The following errors can occur when a request fails:

Error code	Error message	Reason
NotFound	Model: xxx not found!	You are creating a deployment task with a model that does not exist. You are querying, updating, or deleting a deployment task with a model that does not exist.
Conflict	Deployed model xxx already exists, please specify a suffix.	You are creating a deployment task with a suffix that is already in use.
InvalidParameter	Invalid capacity (xx), capacity must be larger than or equal to 0 and multiples of 1 and less than 1000!	You are creating or updating a deployment task with an invalid number of capacity units.

Next steps

When the deployment status is RUNNING, you can call the model for inference. To call a video generation model, see Call the model to generate a video or Call the model to generate an image.