Model compression - Alibaba Cloud Model Studio

Overview

The Model Compression API compresses custom full-parameter fine-tuned models through techniques such as quantization, reducing inference memory footprint and improving throughput. Currently, the compression feature only supports quantization and covers the complete lifecycle from querying templates, creating tasks, polling status, retrieving logs, to canceling/deleting tasks. Typical flow:

List quantizable models and configuration templates → obtain the template_id and the quantizable model
Create a compression task → obtain the job_id
Poll Query compression task / Get compression task logs → until SUCCEEDED / FAILED / CANCELED
After SUCCEEDED, use quantized_output to create a deployment; when no longer needed, Cancel compression task / Delete compression task

Domain for all API endpoints: https://dashscope-intl.aliyuncs.com. Authentication uniformly uses Authorization: Bearer ${YOUR_API_KEY}, and POST requests must include Content-Type: application/json. For the meanings of task object fields and the state machine, see Compression task object; for unified error codes, see Error codes at the end of the document.

Quick Start

The model compression API is currently available only in the Singapore Region. If you use another Region, complete model compression operations through the Bailian console of that Region.

The model compression API provides a complete set of RESTful interfaces covering query templates, creating jobs, polling status, obtaining logs, and canceling/deleting jobs. This document is intended for developers to integrate compression capabilities via OpenAPI or SDK. For console introductions, see related documents.

Prerequisites

Before calling the interfaces in this document, please complete the following:

Activated Alibaba Cloud Bailian service and completed real-name verification.
The current workspace has at least one custom full-parameter fine-tuned model based on qwen3.5-flash-2026-02-23 (completed via the fine-tuning job interface). The current compression feature only supports this model; LoRA models and already-quantized models are not supported.
Obtained an API Key (see Obtain API Key).

The deployment unit specifications supported by the compressed output model are determined by the selected quantization template, and the deployment quantity is configured in the Bailian console under "Model Deployment". The current compression feature is free for a limited time.

Interface List

All interface domains: https://dashscope-intl.aliyuncs.com

#	Method	Path	Description
1	GET	`/api/v1/fine-tunes/compress/templates`	List quantizable models and configuration templates
2	POST	`/api/v1/fine-tunes/compress/jobs`	Create compression job
3	GET	`/api/v1/fine-tunes/compress/jobs`	List compression jobs
4	GET	`/api/v1/fine-tunes/compress/jobs/{job_id}`	Query compression job details
5	GET	`/api/v1/fine-tunes/compress/jobs/{job_id}/logs`	Get compression job logs
6	POST	`/api/v1/fine-tunes/compress/jobs/{job_id}/cancel`	Cancel compression job
7	DELETE	`/api/v1/fine-tunes/compress/jobs/{job_id}`	Delete compression job

Authentication

All interfaces carry the API Key via HTTP Header:

Authorization: Bearer ${YOUR_API_KEY}

Content-Type: application/json applies to POST request body scenarios.

Get Started in 5 Minutes

HTTP

Install the requests library:

pip install requests

Complete example of creating a job, polling, and obtaining the output model:

import requests, time

API_KEY = "YOUR_API_KEY"
BASE = "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress"
HEADERS = {"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"}

# 1. Create compression job
resp = requests.post(f"{BASE}/jobs", headers=HEADERS, json={
    "model": "qwen3.5-flash-2026-02-23-ft-***",  # Custom fine-tuned model ID
    "template_id": "quant-flash-nvfp4-mlp-nomtp",          # Obtain via GET /templates
    "output_model_suffix": "test",                          # Max 8 characters, lowercase letters and digits only
}).json()
job_id = resp["output"]["job_id"]
print("Job:", job_id)

# 2. Poll until terminal state
status = resp["output"]["status"]
while status not in ("SUCCEEDED", "FAILED", "CANCELED"):
    time.sleep(30)
    resp = requests.get(f"{BASE}/jobs/{job_id}", headers=HEADERS).json()
    status = resp["output"]["status"]
    print("Status:", status)

# 3. Process results
if status == "SUCCEEDED":
    print("Quantized model:", resp["output"]["quantized_output"])
    # Use quantized_output to call the model deployment interface for deployment
elif status == "FAILED":
    print("Error:", resp["output"]["error"])

List jobs and obtain logs:

# List all successfully compressed jobs
resp = requests.get(f"{BASE}/jobs", headers=HEADERS, params={"status": "SUCCEEDED", "page_size": 20}).json()
for j in resp["output"]["jobs"]:
    print(j["job_id"], j["template_name"], j["quantized_output"])

# Get job logs
resp = requests.get(f"{BASE}/jobs/{job_id}/logs", headers=HEADERS, params={"offset": 0, "line": 100}).json()
for line in resp["output"]["logs"]:
    print(line)

Output model naming rules:

quantized_output = {base_model}-{output_model_suffix}-{job_id}

For example, base_model is qwen3.5-flash-2026-02-23, suffix is test, job_id is quant-202604111200-a1b2, and the output model ID is:

qwen3.5-flash-2026-02-23-test-quant-202604111200-a1b2

See each interface detail page for cURL usage of each interface.

Compression job object

The model compression API is currently available only in the Singapore Region. If you use another Region, complete model compression in that Region's Bailian console.

Object properties

Response parameters

Field	Type	Description
job_id	String	Job ID
job_name	String	Job name
job_description	String	Job description
status	String	Job status (see Job status)
model	String	Source model ID
base_model	String	Base model ID
template_id	String	ID of the compression template in use
template_name	String	Template name
template_description	String	Template description
training_type	String	Job type, fixed as `quantization`
compress_type	String	Compression type, same as `training_type`, fixed as `quantization`
hyper_parameters	Object	Actually effective hyperparameters (only returns user-visible parameters)
custom_calibration_file_ids	Array<String>	List of file IDs for the custom calibration dataset
quantized_output	String	Model ID produced after quantization (only has a value when SUCCEEDED)
create_time	String	Job creation time
start_time	String	Job start execution time (null when PENDING/QUEUING)
end_time	String	Job completion time (has a value in terminal states)
error	Object	Error information on failure, containing `code` and `message`; null on success
group	String	Job group, fixed as `quantization`
usage	Integer	GPU duration (seconds), appears when SUCCEEDED or CANCELED

Job status

Status	Description
`PENDING`	Job created, waiting for scheduling
`QUEUING`	Entered the scheduling queue, waiting for GPU resources
`RUNNING`	Job in progress
`CANCELING`	Cancel initiated, waiting for termination
`SUCCEEDED`	Job succeeded, the `quantized_output` field returns the produced model ID
`FAILED`	Job failed, the `error.code` / `error.message` fields return the reason
`CANCELED`	Job canceled

Status transition:

PENDING ─→ QUEUING ─→ RUNNING ─→ SUCCEEDED
                  │           │
                  ↓           ↓
                CANCELING ─→ FAILED / CANCELED

List quantizable models and configuration templates

Lists all quantizable custom fine-tuned models of the current user, as well as the compression templates bound to each model. Templates are bound to models; different combinations of model architecture × precision × target MU specifications correspond to different templates.

Only returns custom models that the current user has fully fine-tuned (SFT/DPO/CPT) based on a base model. LoRA fine-tuned models and already-quantized models will not appear in the results.

Endpoint

GET /api/v1/fine-tunes/compress/templates

Request parameters

Parameter	Type	Required	Default	Description
model	String	No	-	Filter by model ID; when a base model name is passed, returns all custom models based on that base model
lang	String	No	zh-CN	Response language: `zh-CN` / `en-US` (see Multi-language support)

Request example

curl "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress/templates" \
  -H "Authorization: Bearer ${API_KEY}"

Response example (minimal)

{
  "request_id": "uuid-string",
  "output": {
    "base_models": ["qwen3.5-flash-2026-02-23"],
    "custom_models": [
      {
        "model": "qwen3.5-flash-2026-02-23-ft-***",
        "model_name": "My SFT fine-tuned model",
        "base_model": "qwen3.5-flash-2026-02-23",
        "templates": [
          {
            "template_id": "quant-flash-nvfp4-mlp-nomtp",
            "template_name": "W4A4 NVFP4 high-performance compression-MU5/MU8/MU9",
            "description": "Balances high precision and high performance under lower-bit compression, further reducing memory usage and improving inference throughput.",
            "compress_type": "quantization",
            "hyper_parameters": []
          }
        ]
      }
    ]
  }
}

Response example (complete: with tunable hyperparameters)

{
  "request_id": "uuid-string",
  "output": {
    "base_models": ["qwen3.5-flash-2026-02-23"],
    "custom_models": [
      {
        "model": "qwen3.5-flash-2026-02-23-ft-***",
        "model_name": "My SFT fine-tuned model",
        "base_model": "qwen3.5-flash-2026-02-23",
        "templates": [
          {
            "template_id": "quant-flash-nvfp4-mlp-nomtp",
            "template_name": "W4A4 NVFP4 high-performance compression-MU5/MU8/MU9",
            "description": "Balances high precision and high performance under lower-bit compression, further reducing memory usage and improving inference throughput.",
            "compress_type": "quantization",
            "hyper_parameters": [
              {
                "name": "calib_input",
                "type": "string",
                "display_name": "Calibration input",
                "description": "Whether to enable calibration input",
                "support_values": ["true"],
                "defaultValue": "true",
                "recommend_value": "true",
                "required": false
              }
            ]
          }
        ]
      }
    ]
  }
}

Response parameters

Field	Type	Description
base_models	Array<String>	List of base model names that support compression
custom_models[].model	String	Model ID
custom_models[].model_name	String	Model display name
custom_models[].base_model	String	Base model name
custom_models[].templates	Array	List of compression configuration templates supported by this model, inherited from its base model's templates
templates[].template_id	String	Template ID, passed as the template_id parameter when creating a compression task
templates[].template_name	String	Template name (supports multi-language; returns the corresponding language version based on the `lang` parameter)
templates[].description	String	Template description (supports multi-language; returns the corresponding language version based on the `lang` parameter)
templates[].compress_type	String	Compression type, fixed as `quantization`
templates[].hyper_parameters	Array	Tunable hyperparameters; an empty array means no tunable hyperparameters
hyper_parameters[].name	String	Parameter name (used as the Key when creating a task)
hyper_parameters[].type	String	Type: `number` (numeric, used with `data_range`/`step`) / `string` (enumeration, used with `support_values`)
hyper_parameters[].display_name	String	Parameter display name (supports multi-language; returns the corresponding language version based on the `lang` parameter)
hyper_parameters[].description	String	Parameter description (supports multi-language; returns the corresponding language version based on the `lang` parameter)
hyper_parameters[].defaultValue	String	Default value
hyper_parameters[].recommend_value	String	Recommended value
hyper_parameters[].required	Boolean	Whether required
hyper_parameters[].support_values	Array<String>	List of enumeration values (only present when `type=string`), e.g. `["instruct", "think", "hybrid"]`
hyper_parameters[].data_range	Array<String>	Numeric range (only present when `type=number`), e.g. `["64","256"]` indicates a range of 64~256
hyper_parameters[].step	Integer	Step size (only present when `type=number`)

Create a compression job

Endpoint

POST /api/v1/fine-tunes/compress/jobs

Request parameters

Parameter	Type	Required	Default	Description
model	String	Yes	-	Source model ID, which can be obtained via the API
template_id	String	Yes	-	Compression template ID, which can be obtained via the API
job_name	String	No	Auto-generated	Job name; duplicates are not allowed under the same user; up to 50 characters
job_description	String	No	-	Job description; up to 200 characters
hyper_parameters	Object	No	Template default value	Hyperparameter overrides (key-value); only pass the items you want to override
custom_calibration_file_ids	Array<String>	No	-	List of custom calibration dataset file IDs (dataset group ID, in the format `file-{32hex}`); takes effect only when `calib_input=true`. The dataset must be created and published in Data Management first
output_model_suffix	String	No	-	Suffix of the quantized output model name; up to 8 characters, only lowercase letters and digits. Output model name format: `{base_model}-{suffix}-{job_id}`

Request example

curl -X POST "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress/jobs" \
  -H "Authorization: Bearer ${API_KEY}" \
  -H "Content-Type: application/json" \
  -d '{
    "job_name": "qwen3.5-flash compression job",
    "model": "qwen3.5-flash-2026-02-23-ft-***",
    "template_id": "quant-flash-nvfp4-mlp-nomtp",
    "custom_calibration_file_ids": ["file-***"],
    "output_model_suffix": "test"
  }'

Response example

{
  "request_id": "uuid-string",
  "output": {
    "job_id": "quant-202604111200-a1b2",
    "job_name": "qwen3.5-flash compression job",
    "status": "PENDING",
    "model": "qwen3.5-flash-2026-02-23-ft-***",
    "base_model": "qwen3.5-flash-2026-02-23",
    "template_id": "quant-flash-nvfp4-mlp-nomtp",
    "template_name": "W4A4 NVFP4 high-performance compression-MU5/MU8/MU9",
    "training_type": "quantization",
    "compress_type": "quantization",
    "hyper_parameters": {},
    "custom_calibration_file_ids": ["file-***"],
    "quantized_output": null,
    "create_time": "2026-04-11 12:00:00",
    "start_time": null,
    "end_time": null,
    "error": null,
    "group": "quantization"
  }
}

Response parameters: Field meanings are the same as the request parameters.

List compression jobs

Supports filtering by status, model, template, quantization spec, algorithm, time range, job name/ID, and supports sorting by creation time and pagination. Endpoint

GET /api/v1/fine-tunes/compress/jobs

Request parameters

Parameter	Type	Required	Default	Description
status	String	No	-	Filter by status (e.g., RUNNING, SUCCEEDED)
model	String	No	-	Filter by source model ID
template_id	String	No	-	Filter by template ID
quant_spec	String	No	-	Filter by quantization spec (e.g., `w4a16`, `w8a8`)
quant_method	String	No	-	Filter by quantization algorithm (e.g., `gptq`, `awq`, `fp8`)
start_time	String	No	-	Job start time is no earlier than this value. Format: `yyyy-MM-dd HH:mm:ss` / ISO-8601 / `yyyy-MM-dd`
end_time	String	No	-	Job end time is no later than this value, format same as `start_time`
job_name	String	No	-	Fuzzy match by job name
job_id	String	No	-	Fuzzy match by job ID
search_key	String	No	-	Search keyword. When `select_key` is not provided, fuzzy matches both `job_id` and `job_name`; when combined with `select_key`, only the specified field is matched
select_key	String	No	-	Search field for `search_key`, options: `job_id` / `job_name`
sort_by	String	No	create_time	Sort field, currently only supports `create_time`
sort_order	String	No	desc	Sort direction, `asc` ascending / `desc` descending
page_no	Integer	No	1	Page number
page_size	Integer	No	10	Page size, max 100

Request examples

# Combined search: by status + algorithm + pagination
curl "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress/jobs?status=SUCCEEDED&quant_method=gptq&page_size=10" \
  -H "Authorization: Bearer ${API_KEY}"

# Time range + search by job name + ascending by create time
curl "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress/jobs?start_time=2026-04-01&end_time=2026-04-30&search_key=qwen3&select_key=job_name&sort_by=create_time&sort_order=asc" \
  -H "Authorization: Bearer ${API_KEY}"

Response example

{
  "request_id": "uuid-string",
  "output": {
    "total": 42,
    "page_no": 1,
    "page_size": 10,
    "jobs": [
      {
        "job_id": "quant-202604111200-a1b2",
        "job_name": "qwen3.5-flash compression job",
        "status": "SUCCEEDED",
        "model": "qwen3.5-flash-2026-02-23-ft-***",
        "base_model": "qwen3.5-flash-2026-02-23",
        "template_id": "quant-flash-nvfp4-mlp-nomtp",
        "template_name": "W4A4 NVFP4 high-performance compression-MU5/MU8/MU9",
        "training_type": "quantization",
        "compress_type": "quantization",
        "custom_calibration_file_ids": ["file-***"],
        "quantized_output": "qwen3.5-flash-2026-02-23-test-quant-202604111200-a1b2",
        "create_time": "2026-04-11 12:00:00",
        "start_time": "2026-04-11 12:02:30",
        "end_time": "2026-04-11 13:02:30",
        "group": "quantization",
        "usage": 3600
      }
    ]
  }
}

Response parameters

Field	Type	Description
total	Integer	Total number of matching jobs
page_no	Integer	Current page number
page_size	Integer	Page size
jobs	Array	Job list, field meanings are the same as the response parameters of creating a compression job

Query a compression job

Endpoint

GET /api/v1/fine-tunes/compress/jobs/{job_id}

Request example

curl "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress/jobs/quant-202604111200-a1b2" \
  -H "Authorization: Bearer ${API_KEY}"

Response example

{
  "request_id": "uuid-string",
  "output": {
    "job_id": "quant-202604111200-a1b2",
    "job_name": "qwen3.5-flash compression job",
    "job_description": "...",
    "status": "SUCCEEDED",
    "model": "qwen3.5-flash-2026-02-23-ft-***",
    "base_model": "qwen3.5-flash-2026-02-23",
    "template_id": "quant-flash-nvfp4-mlp-nomtp",
    "template_name": "W4A4 NVFP4 high-performance compression-MU5/MU8/MU9",
    "template_description": "Maintains high accuracy and high performance under lower-bit compression, further reducing GPU memory usage and improving inference throughput.",
    "training_type": "quantization",
    "compress_type": "quantization",
    "hyper_parameters": {},
    "custom_calibration_file_ids": ["file-***"],
    "quantized_output": "qwen3.5-flash-2026-02-23-test-quant-202604111200-a1b2",
    "create_time": "2026-04-11 12:00:00",
    "start_time": "2026-04-11 12:02:30",
    "end_time": "2026-04-11 13:02:30",
    "error": null,
    "group": "quantization",
    "usage": 3600
  }
}

Response parameters

Field	Type	Description
job_id	String	Job ID, which can be obtained via the create compression job or list compression jobs interface
job_name	String	Job name
job_description	String	Job description
status	String	Job status (see Job status for details)
model	String	Source model ID
base_model	String	Base model ID
template_id	String	ID of the compression template used
template_name	String	Template name
template_description	String	Template description
training_type	String	Job type, fixed as `quantization`
compress_type	String	Compression type, same as `training_type`, fixed as `quantization`
hyper_parameters	Object	Actually effective hyperparameters (only returns user-visible parameters)
custom_calibration_file_ids	Array<String>	List of file IDs for the custom calibration dataset
quantized_output	String	Model ID produced after quantization (only has a value when SUCCEEDED), can be used by the Create deployment interface for model deployment
create_time	String	Job creation time
start_time	String	Job start execution time (null when PENDING/QUEUING)
end_time	String	Job completion time (has a value at terminal states)
error	Object	Error information on failure, containing `code` and `message`; null on success
group	String	Job group, fixed as `quantization`
usage	Integer	GPU duration (seconds), present when SUCCEEDED or CANCELED

Get compression job logs

Endpoint

GET /api/v1/fine-tunes/compress/jobs/{job_id}/logs

Where {job_id} is the compression job ID, which can be obtained via the create compression job or list compression jobs interface. Request parameters

Parameter	Type	Required	Default	Description
offset	Integer	No	0	Skip the first N lines and start reading from the (N+1)th line
line	Integer	No	100	Number of lines to read, up to 1000

Request example

curl "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress/jobs/quant-202604111200-a1b2/logs?offset=0&line=50" \
	-H "Authorization: Bearer ${API_KEY}"

Response example

{
  "request_id": "uuid-string",
  "output": {
    "total": 15,
    "logs": [
      "2026-04-11 12:02:35 - INFO - Starting quantization...",
      "2026-04-11 12:30:00 - INFO - Quantization progress: 100%",
      "2026-04-11 12:35:00 - INFO - Quantization succeeded!"
    ]
  }
}

Log display rules

When a custom calibration dataset is provided for the job, the logs contain a data processing completion marker data process succeeded, start to quantization
The log interface has filtered out internal system markers and only returns user-readable compression progress information

Cancel a compression job

You can cancel only jobs in the PENDING, QUEUING, or RUNNING state. Cancellation is an asynchronous operation. The job first enters the CANCELING transitional state and eventually becomes CANCELED. Endpoint

POST /api/v1/fine-tunes/compress/jobs/{job_id}/cancel

Where {job_id} is the compression job ID, which can be obtained from the Create compression job or List compression jobs API. Request example

curl -X POST "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress/jobs/quant-202604111200-a1b2/cancel" \
  -H "Authorization: Bearer ${API_KEY}"

Response example

{
  "request_id": "uuid-string",
  "output": { "status": "success" }
}

Delete a compression job

You can delete only jobs in a terminal state (SUCCEEDED / FAILED / CANCELED). Deleting a job record does not delete the already-generated quantized model (quantized_output). Endpoint

DELETE /api/v1/fine-tunes/compress/jobs/{job_id}

Where {job_id} is the compression job ID, which can be obtained from the Create compression job or List compression jobs API. Request example

curl -X DELETE "https://dashscope-intl.aliyuncs.com/api/v1/fine-tunes/compress/jobs/quant-202604111200-a1b2" \
  -H "Authorization: Bearer ${API_KEY}"

Response example

{
  "request_id": "uuid-string",
  "output": { "status": "success" }
}

Error code

Common error codes

Error code	HTTP	Description
`InvalidParameter`	400	Invalid request parameter
`MissingParameter`	400	Missing required parameter
`Unauthorized`	401	Authentication failed
`Forbidden`	403	No permission to access
`ResourceNotFound`	404	Resource does not exist
`UnsupportedOperation`	400	The resource state does not allow this operation (e.g., canceling a job that is already in a terminal state)
`QuotaExceeded`	429	Quota exceeded
`InternalError`	500	Internal service error

Business error codes

The following business error codes are listed by scenario. External Code is the actual code field value returned by the interface. Parameter validation

External Code	HTTP	Description
`InvalidParameter`	400	Missing required parameter `model`
`InvalidParameter`	400	Missing required parameter `template_id`
`InvalidParameter`	400	Direct quantization of base models is not supported
`InvalidParameter`	400	The specified configuration template does not exist
`InvalidParameter`	400	The current model does not support this compression template
`InvalidParameter`	400	The model does not support quantization
`InvalidParameter`	400	LoRA fine-tuned models do not support quantization
`InvalidParameter`	400	Model data unavailable
`InvalidParameter`	400	The job name contains unsupported characters
`InvalidParameter`	400	`output_model_suffix` exceeds 8 characters
`InvalidParameter`	400	The source model is not ready
`AccessDenied`	403	No permission to use this compression template

Hyperparameter validation

External Code	HTTP	Description
`InvalidParameter`	400	Required hyperparameter not provided
`InvalidParameter`	400	Unknown hyperparameter provided
`InvalidParameter`	400	The hyperparameter value is not in the enumeration list
`InvalidParameter`	400	The hyperparameter value is out of range
`InvalidParameter`	400	The hyperparameter value is not a valid number

Job query

External Code	HTTP	Description
`NotFound`	404	The specified compression job does not exist
`InvalidParameter`	400	Missing required parameter `job_id`

Pagination and time parameters

External Code	HTTP	Description
`InvalidParameter`	400	Invalid page number parameter (must be ≥ 1)
`InvalidParameter`	400	Invalid page size (must be 1–100)
`InvalidParameter`	400	Invalid time format

Error response example

{
  "request_id": "uuid-string",
  "code": "InvalidParameter",
  "message": "The specified model 'xxx-lora-yyy' is a LoRA model and not supported for quantization."
}

​Overview

​Quick Start

​Prerequisites

​Interface List

​Authentication

​Get Started in 5 Minutes

​Compression job object

​Object properties

​Job status

​List quantizable models and configuration templates

​Create a compression job

​List compression jobs

​Query a compression job

​Get compression job logs

​Cancel a compression job

​Delete a compression job

​Error code

​Common error codes

​Business error codes

​Error response example

Overview

Quick Start

Prerequisites

Interface List

Authentication

Get Started in 5 Minutes

Compression job object

Object properties

Job status

List quantizable models and configuration templates

Create a compression job

List compression jobs

Query a compression job

Get compression job logs

Cancel a compression job

Delete a compression job

Error code

Common error codes

Business error codes

Error response example