Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Alibaba Cloud / Tongyi Embedding Vision

Tongyi Embedding Vision Plus · March 2026

The March vision embedding release adds fused vectors, configurable dimensions and resolution controls.

availableClosed weightsEmbeddings
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
27 of 33 specification fields collected
Input token limit1K1,024 tokens
Vector dimensions1152, 1024, 512, 256, 128, 64See specification notes below
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
1,024 tokensSource for Maximum input tokens (opens in a new tab)

Per text input; not an aggregate multimodal request ceiling.

Maximum output
Not collected
Input modalities
Text, Image, VideoSource for Input modalities (opens in a new tab)
Reasoning
Not collected
Streaming
Not collected
Knowledge cutoff
Not collected
Release date
2026-03-20Source for Release date (opens in a new tab)

Earliest reviewed regional release; not inferred from a snapshot suffix.

Parameters
Not collected
Architecture
Built on the Qwen3 foundation; private architecture details not disclosedSource for Architecture (opens in a new tab)
Languages
View all 33 entries
  • Chinese
  • Japanese
  • Korean
  • Indonesian
  • Vietnamese
  • Thai
  • English
  • French
  • German
  • Russian
  • Portuguese
  • Spanish
  • Italian
  • Swedish
  • Danish
  • Czech
  • Norwegian
  • Dutch
  • Finnish
  • Turkish
  • Polish
  • Swahili
  • Romanian
  • Serbian
  • Greek
  • Kazakh
  • Uzbek
  • Cebuano
  • Arabic
  • Urdu
  • Persian
  • Hindi / Devanagari
  • Hebrew
Source for Languages (opens in a new tab)
API aliases
tongyi-embedding-vision-plus-2026-03-06Source for API aliases (opens in a new tab)

Pinned and rolling selectors are separate records; equivalence is not asserted.

Embedding dimensions
1152, 1024, 512, 256, 128, 64Source for Embedding dimensions (opens in a new tab)
Embedding output modes
Independent, FusedSource for Embedding output modes (opens in a new tab)
Multimodal fusion
Put text, image and video in the same content object. Do not use enable_fusion, which belongs to Qwen3-VL.Source for Multimodal fusion (opens in a new tab)
Embedding input types
At most 20 text elements, 64 images and 8 videos per request; multi_images sequences are supported.Source for Embedding input types (opens in a new tab)
Image input limits
Recommended at most 5 MB per image; maximum 10 MB; at most 64 imagesSource for Image input limits (opens in a new tab)
Source video limits
At most 50 MB per video; H.264 or H.265 encoding; at most 8 videosSource for Source video limits (opens in a new tab)
Image input formats
JPEG, PNG, WEBP, BMP, TIFF, ICO, DIB, ICNS, SGISource for Image input formats (opens in a new tab)
Video input formats
MP4, MPEG, MOV, MPG, WEBM, AVI, FLV, MKVSource for Video input formats (opens in a new tab)
Video frame sampling
fps 0–1 (default 1); max_video_frames at most 64 (default 8)Source for Video frame sampling (opens in a new tab)
Image token calculation
res_level 0/1/2/3 uses 127/402/578/1,026 tokens per image; default level 1Source for Image token calculation (opens in a new tab)
Request mode
DashScope multimodal embeddings; public-URL/Base64 images and public-URL videosSource for Request mode (opens in a new tab)
Request quotas
1,200 RPM; 9,600,000 TPMSource for Request quotas (opens in a new tab)

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
tongyi-embedding-vision-plus-2026-03-06
Access
https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding. Use a Beijing workspace API key.
Regions
China (Beijing)
Request quotas
1,200 RPM; 9,600,000 TPMSource for Request quotas (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
EmbeddingsImage/video input1CN¥0.501M tokens
EmbeddingsText input1CN¥0.501M tokens

Rate conditions

  1. Published domestic-documentation CNY tariff, not a currency conversion from USD. Beijing workspace. No generated-answer estimate is inferred.
Pricing source (opens in a new tab)

Data & enterprise terms

Account-specific regional retention and enterprise policy require confirmation.

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 provider names this native API release separately from public Qwen3/GTE/GME/GVE checkpoints, and no exact release was found in the reviewed official Qwen or Alibaba-NLP weight inventories. Absence is supporting evidence, not proof of private checkpoint lineage. Base-model parameter counts are not transferred to this service.

Classification evidence (opens in a new tab)

Lifecycle dates

Platform release · China (Beijing)20 Mar 2026Source (opens in a new tab)

Sources reviewed