Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Alibaba Cloud / Qwen2.5

Qwen2.5-Max

API-served mixture-of-experts model for text generation, code and multilingual tasks.

Language models (LLMs)
availableClosed weightsText generationCodingAgents
Official documentation (opens in a new tab)Reviewed 12 Oct 2026
16 of 20 specification fields collected
Input token limit30.7K30,720 tokens
Maximum output8.2K8,192 tokens
Input / 1M tokens$1.60Alibaba Model Studio · Singapore · International
Output / 1M tokens$6.40USD · observed 12 Oct 2026

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
32,768 tokensSource for Context / input limit (opens in a new tab)
Maximum output
8,192 tokensSource for Maximum output (opens in a new tab)
Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
TextSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
SupportedSource for Tool calling (opens in a new tab)

Available on at least one reviewed endpoint when true. Regional differences appear under access and pricing.

Structured output
SupportedSource for Structured output (opens in a new tab)

Available on at least one reviewed endpoint when true. Regional differences appear under access and pricing.

Streaming
SupportedSource for Streaming (opens in a new tab)

OpenAI-compatible Chat Completions streaming; endpoint-specific availability should be checked before deployment.

Batch processing
SupportedSource for Batch processing (opens in a new tab)

Available on at least one reviewed endpoint when true. Regional differences appear under access and pricing.

Fine-tuning
Not supportedSource for Fine-tuning (opens in a new tab)

Available on at least one reviewed endpoint when true. Regional differences appear under access and pricing.

Knowledge cutoff
Not collected
Release date
2025-01-28Source for Release date (opens in a new tab)
Parameters
Not collected
Architecture
Mixture of expertsSource for Architecture (opens in a new tab)
Languages
Not collected
Context caching
SupportedSource for Context caching (opens in a new tab)

Available on at least one reviewed endpoint when true. Regional differences appear under access and pricing.

Maximum input tokens
30,720 tokensSource for Maximum input tokens (opens in a new tab)
API aliases
qwen-max, qwen-max-2025-01-25Source for API aliases (opens in a new tab)
Published training data
More than 20 trillion pretraining tokensSource for Published training data (opens in a new tab)
Search grounding
SupportedSource for Search grounding (opens in a new tab)

Available on at least one reviewed endpoint when true. Regional differences appear under access and pricing.

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Alibaba Model Studio · Singapore · International

Provider source (opens in a new tab)
API model ID
qwen-max
Access
OpenAI-compatible Chat Completions or DashScope, with a regional workspace API key. Listed scope describes this price/capability configuration; account eligibility must be checked.
Regions
Singapore / International deployment scope
Tool calling
Not supportedSource for Tool calling (opens in a new tab)
Structured output
SupportedSource for Structured output (opens in a new tab)
Search grounding
Not supportedSource for Search grounding (opens in a new tab)
Context caching
SupportedSource for Context caching (opens in a new tab)
Batch processing
SupportedSource for Batch processing (opens in a new tab)
Fine-tuning
Not supportedSource for Fine-tuning (opens in a new tab)
Published request quota
600 requests/minuteSource for Published request quota (opens in a new tab)
Published token quota
1,000,000 tokens/minuteSource for Published token quota (opens in a new tab)
INPUT / 1M TOKENS$1.60
OUTPUT / 1M TOKENS$6.40

USD list rates for Singapore / International. Cached input means implicit cache reads. Separate explicit-cache and batch charges are listed below. Published card prices are retained without currency conversion. Cached input price: $0.32 / 1M tokens (base tier).

USD · observed 12 Oct 2026 · Pricing evidence (opens in a new tab)
Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
BatchInput(Batch File)1$0.801M tokens
BatchOutput(Batch File)1$3.201M tokens

Rate conditions

  1. Alternative billing mode for this same endpoint. Explicit creation replaces normal input billing for written tokens; explicit reads replace input billing for matching tokens. Batch prices are separate and do not stack with cache discounts.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Alibaba Model Studio · China (Beijing)

Provider source (opens in a new tab)
API model ID
qwen-max
Access
OpenAI-compatible Chat Completions or DashScope, with a regional workspace API key. Listed scope describes this price/capability configuration; account eligibility must be checked.
Regions
China (Beijing)
Tool calling
SupportedSource for Tool calling (opens in a new tab)
Structured output
SupportedSource for Structured output (opens in a new tab)
Search grounding
SupportedSource for Search grounding (opens in a new tab)
Context caching
SupportedSource for Context caching (opens in a new tab)
Batch processing
SupportedSource for Batch processing (opens in a new tab)
Fine-tuning
Not supportedSource for Fine-tuning (opens in a new tab)
Published request quota
1200 requests/minuteSource for Published request quota (opens in a new tab)
Published token quota
1,000,000 tokens/minuteSource for Published token quota (opens in a new tab)
INPUT / 1M TOKENS$0.345
OUTPUT / 1M TOKENS$1.377

USD list rates for China (Beijing). Cached input means implicit cache reads. Separate explicit-cache and batch charges are listed below. Published card prices are retained without currency conversion. Cached input price: $0.072 / 1M tokens (base tier).

USD · observed 12 Oct 2026 · Pricing evidence (opens in a new tab)
Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
BatchInput(Batch File)1$0.1721M tokens
BatchOutput(Batch File)1$0.6881M tokens

Rate conditions

  1. Alternative billing mode for this same endpoint. Explicit creation replaces normal input billing for written tokens; explicit reads replace input billing for matching tokens. Batch prices are separate and do not stack with cache discounts.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

Alibaba Model Studio · Singapore · International

qwen-max
Documented
Support evidence (opens in a new tab)
How it works
Automatic implicit cacheSource for How it works (opens in a new tab)
Minimum reusable input
1,024 tokensSource for Minimum reusable input (opens in a new tab)
Lifetime & refresh
Implicit lifetime is not fixed; unused cache data is periodically evicted.Source for Lifetime & refresh (opens in a new tab)
How to enable it
Reuse an identical prompt prefix. Implicit caching is automatic; a hit is not guaranteed.Source for How to enable it (opens in a new tab)
Usage fields
Inspect the API usage cache fields; field names depend on API, region and model.Source for Usage fields (opens in a new tab)
Billing rules
Cache-hit input uses the implicit cache rate; uncached input and output retain standard rates.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Explicit and implicit modes are mutually exclusive. Batch and cache discounts do not stack. Identical prefixes and eligible deployment scope are required.Source for Limits & data boundaries (opens in a new tab)
Cache reads, writes & storage
ChargeApplies toRate / unit
Cache read1Text tokens · Base tier$0.32 / 1M tokensPricing evidence (opens in a new tab)

Rate conditions

  1. USD list rates for Singapore / International. Cached input means implicit cache reads. Separate explicit-cache and batch charges are listed below. Published card prices are retained without currency conversion.

Observed 12 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.

Caching reviewed 12 Oct 2026

Alibaba Model Studio · China (Beijing)

qwen-max
Documented
Support evidence (opens in a new tab)
How it works
Automatic implicit cacheSource for How it works (opens in a new tab)
Minimum reusable input
1,024 tokensSource for Minimum reusable input (opens in a new tab)
Lifetime & refresh
Implicit lifetime is not fixed; unused cache data is periodically evicted.Source for Lifetime & refresh (opens in a new tab)
How to enable it
Reuse an identical prompt prefix. Implicit caching is automatic; a hit is not guaranteed.Source for How to enable it (opens in a new tab)
Usage fields
Inspect the API usage cache fields; field names depend on API, region and model.Source for Usage fields (opens in a new tab)
Billing rules
Cache-hit input uses the implicit cache rate; uncached input and output retain standard rates.Source for Billing rules (opens in a new tab)
Limits & data boundaries
Explicit and implicit modes are mutually exclusive. Batch and cache discounts do not stack. Identical prefixes and eligible deployment scope are required.Source for Limits & data boundaries (opens in a new tab)
Cache reads, writes & storage
ChargeApplies toRate / unit
Cache read1Text tokens · Base tier$0.072 / 1M tokensPricing evidence (opens in a new tab)

Rate conditions

  1. USD list rates for China (Beijing). Cached input means implicit cache reads. Separate explicit-cache and batch charges are listed below. Published card prices are retained without currency conversion.

Observed 12 Oct 2026. Write rates and storage fees are different charges; use the billing rules above.

Caching reviewed 12 Oct 2026

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Inference from the official API-only Qwen2.5-Max release, the later announcement that Qwen3.8 is the first Max-class public-weight release, and the reviewed official public-weight inventory. No claim that every Qwen-Max-class release is closed.

Classification evidence (opens in a new tab)

Lifecycle dates

Released28 Jan 2025Source (opens in a new tab)

Sources reviewed