BGE Base Zh is an open-source embedding model created by BAAI. It is designed to map Chinese text into dense vector representations. The model is built for tasks like retrieval, classification, clustering, and semantic search.
Input
Output
Context
512
Max Output
n/a
Parameters
102.3M
Input Modalities
Output Modalities
Some values come from web research and may be incorrect. Report incorrect data
None the catalog can show from stated facts or measured scores.
Download 0.41 GB · Runs on vLLM v0.32.0 · 2 GGUF builds
| Precision | Weights | Smallest GPU that fits |
|---|---|---|
| FP16 / BF16 | 0.19 GB | 1x NVIDIA T4 (16 GB) |
| INT8 | 0.10 GB | 1x NVIDIA GeForce RTX 3090 (24 GB) |
| INT4 | 0.05 GB | 1x NVIDIA T4 (16 GB) |
Computed by Inferbase from the 102,268,160 parameters in the creator's Hugging Face repository; weights only, as serving adds the framework and the input being processed.
Public resources
Specification sources
As of Nov 10, 2026.
Answered from the facts on this page, with their sources.
BGE Base Zh has a context window of 512 tokens (source: Web research). A provider may serve less; Inferbase routes on what each provider serves.
Yes. BGE Base Zh is an open-weight model under the MIT license. You can download and self-host it within the license terms.
BGE Base Zh has 102,268,160 parameters (102.3M), as stated in the creator's Hugging Face repository.
Strengths: Open weights under a permissive license (mit). Each is computed from the model's stated facts and benchmark results.
BGE Base Zh accepts Text and produces Embedding. The Capabilities card lists the API features it supports, such as function calling and structured output.
No benchmark source has published a result for BGE Base Zh yet, so it is shown as unmeasured, not estimated.
Not on Inferbase Managed today. Add your own provider key or endpoint that serves BGE Base Zh, and Inferbase runs it on your account with the same routing and logs.
The weights take 0.2 GB at FP16 / BF16, 0.1 GB at INT8, 0.0 GB at INT4, computed from the 102,268,160 parameters its creator publishes. Serving adds the framework and the input being processed.
The model card of BGE Base Zh lists one language: Chinese. That is what BAAI states; the catalog does not test language quality.
Start on the free plan with 5,000 requests a month, no card required.