Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Alibaba Cloud / Tongyi Text Embedding

Tongyi Text Embedding v2

Ten-language text embeddings with 2,048 tokens per item and separate synchronous and asynchronous request limits.

Embeddings
availableClosed weightsEmbeddings
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
20 of 28 specification fields collected
Input token limit2K2,048 tokens
Vector dimensions1.5K1,536 dimensions
Input billingSee ratesAlibaba Model Studio · China (Beijing)
Output billingSee ratesPublished units and conditions below

Specifications

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

Context / input limit
Not collected
Maximum input tokens
2,048 tokensSource for Maximum input tokens (opens in a new tab)

Per text item; a synchronous call accepts at most 25 items.

Maximum output
Not collected
Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
EmbeddingsSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not supportedSource for Tool calling (opens in a new tab)
Structured output
Not supportedSource for Structured output (opens in a new tab)
Streaming
Not collected
Batch processing
Not collected

The synchronous model card marks Batch Inference unsupported but publishes a Batch File tariff. Its separate async service is documented; eligibility for the discounted Batch API remains unresolved.

Fine-tuning
Not supportedSource for Fine-tuning (opens in a new tab)
Knowledge cutoff
Not collected
Release date
2024-04-09Source for Release date (opens in a new tab)

Date listed in the regional platform release history, not necessarily the original research launch.

Parameters
Not collected
Architecture
Not collected
Languages
Chinese, English, Spanish, French, Portuguese, Indonesian, Japanese, Korean, German, RussianSource for Languages (opens in a new tab)
Context caching
Not supportedSource for Context caching (opens in a new tab)
Search grounding
Not supportedSource for Search grounding (opens in a new tab)
Prefix completion
Not supportedSource for Prefix completion (opens in a new tab)
API aliases
text-embedding-v2Source for API aliases (opens in a new tab)

The separately listed async selectors are access variants, not guaranteed byte-identical vector aliases.

Request quotas
1,800 RPM; 1,200,000 TPMSource for Request quotas (opens in a new tab)

Provider-published default quota; account-specific limits can differ.

Embedding dimensions
1,536 dimensionsSource for Embedding dimensions (opens in a new tab)
Default vector dimensions
1,536Source for Default vector dimensions (opens in a new tab)
Items per embedding request
25Source for Items per embedding request (opens in a new tab)
Embedding output modes
Dense, fixed-size float vectorsSource for Embedding output modes (opens in a new tab)
Embedding input types
String, string array, or text file; query/document text_type on DashScopeSource for Embedding input types (opens in a new tab)
Request mode
OpenAI-compatible embeddings or DashScope text embedding; asynchronous file endpoint also availableSource for Request mode (opens in a new tab)
Free usage allowance
500,000 tokens; regional eligibility and 90-day expiry applySource for Free usage allowance (opens in a new tab)

Overview starts expiry from the latest activation, release or approval date; API guide says activation. Confirm the account quota.

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 · China (Beijing)

Provider source (opens in a new tab)
API model ID
text-embedding-v2
Access
https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/embeddings. Use a Beijing workspace API key.
Regions
China (Beijing)
Tool calling
Not supportedSource for Tool calling (opens in a new tab)
Structured output
Not supportedSource for Structured output (opens in a new tab)
Search grounding
Not supportedSource for Search grounding (opens in a new tab)
Prefix completion
Not supportedSource for Prefix completion (opens in a new tab)
Context caching
Not supportedSource for Context caching (opens in a new tab)
Batch processing
Not collected

The synchronous model card marks Batch Inference unsupported but publishes a Batch File tariff. Its separate async service is documented; eligibility for the discounted Batch API remains unresolved.

Fine-tuning
Not supportedSource for Fine-tuning (opens in a new tab)
Request quotas
1,800 RPM; 1,200,000 TPMSource for Request quotas (opens in a new tab)

Provider-published default quota; account-specific limits can differ.

Maximum input tokens
2,048 tokensSource for Maximum input tokens (opens in a new tab)

Per text item; a synchronous call accepts at most 25 items.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
EmbeddingsText input1CN¥0.701M tokens
Batch embeddingsBatch text input2CN¥0.351M tokens

Rate conditions

  1. Beijing native CNY tariff. No generated-answer estimate.
  2. Beijing native CNY tariff. Conditional Batch File rate; card marks Batch Inference unsupported. Do not apply this rate to text-embedding-async selectors. No generated-answer estimate.
Pricing source (opens in a new tab)

Data & enterprise terms

Regional retention and enterprise account terms have not been verified.

Alibaba Model Studio · Beijing async files

Provider source (opens in a new tab)
API model ID
text-embedding-async-v2
Access
POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding with X-DashScope-Async: enable and input.url. Poll GET /api/v1/tasks/{task_id}. Grouped as the versioned embedding service batch interface; exact checkpoint equivalence is not asserted.
Regions
China (Beijing)
Tool calling
Not supportedSource for Tool calling (opens in a new tab)
Structured output
Not supportedSource for Structured output (opens in a new tab)
Search grounding
Not supportedSource for Search grounding (opens in a new tab)
Prefix completion
Not supportedSource for Prefix completion (opens in a new tab)
Context caching
Not supportedSource for Context caching (opens in a new tab)
Batch processing
Not collected

Dedicated file task API is supported. The generic Batch Inference capability is marked unsupported by its card; it is not the discounted Batch API.

Fine-tuning
Not supportedSource for Fine-tuning (opens in a new tab)
Request quotas
1 submission/second; at most 50 queued/running tasks, with 3 running concurrentlySource for Request quotas (opens in a new tab)
Maximum input tokens
2,048 tokensSource for Maximum input tokens (opens in a new tab)

Per file line, not total file tokens.

Request mode
Asynchronous file task; X-DashScope-Async: enableSource for Request mode (opens in a new tab)
Items per embedding request
100,000Source for Items per embedding request (opens in a new tab)
Embedding dimensions
1,536 dimensionsSource for Embedding dimensions (opens in a new tab)
Maximum input content size
200 MB per fileSource for Maximum input content size (opens in a new tab)
Task ID lifetime
Task ID, status and result URL retained for 24 hours; download results promptlySource for Task ID lifetime (opens in a new tab)
Free usage allowance
20 million tokens per async selector; 90-day regional trialSource for Free usage allowance (opens in a new tab)

Overview and API guide differ on the start of the trial clock.

Usage accounting
total_tokens counts successfully processed inputSource for Usage accounting (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
Asynchronous embeddingsFile text input1CN¥0.701M tokens

Rate conditions

  1. Dedicated asynchronous file selector. The API and model card agree on CNY 0.7 per million input tokens; the synchronous selector Batch File discount is not applied.
Pricing source (opens in a new tab)

Data & enterprise terms

Task metadata and result URL expire after 24 hours. This does not establish a complete data-retention or training policy.

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.

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

Classification inference: the official developer card distinguishes the commercial text embedding v1/v2/v3 and GTE rerank APIs from public GTE/GME checkpoints. The reviewed public inventory does not establish an exact weight release for this service. Private parameter counts, architecture and checkpoint equivalence remain undisclosed.

Classification evidence (opens in a new tab)

Lifecycle dates

Platform release · China (Beijing)9 Apr 2024Source (opens in a new tab)

Sources reviewed