InferX Beta Serverless GPU Inference Platform, Built for Agent-Native Workloads

ERNIE-4.5-21B-A3B-PT

ERNIE-4.5-21B-A3B-PT is a high-efficiency Large Language Model (LLM) developed by Baidu, released as part of their ERNIE 4.5 family in late 2025. It is specifically designed to balance high-level reasoning capabilities with low computational costs.
baidu text text2text
Log in to deploy: this public page shows the catalog model details, but deployment and customization stay behind login.
Log in to deploy

Metadata

Provider
baidu
Modality
text
API type
text2text
Source
Created
2026-04-12 00:30:55 UTC
Updated
2026-04-13 03:01:13 UTC
Catalog version
2
Visibility
Published

Specifications

Parameters
21.00B
MoE
No
Max model length
2000
Image
vllm/vllm-openai:v0.16.0

Default Deploy Config

GPU count
1
vRAM
70000 MB
Summary
1xGPU 70000 MB

Recommended Use Cases

Model Spec