Baidu: ERNIE 4.5 21B A3B Thinking

Text input Text output
Author's Description

ERNIE-4.5-21B-A3B-Thinking is Baidu's upgraded lightweight MoE model, refined to boost reasoning depth and quality for top-tier performance in logical puzzles, math, science, coding, text generation, and expert-level academic benchmarks.

Key Specifications
Cost
$$$$
Context
131K
Parameters
21B
Released
Oct 09, 2025
Speed
Ability
Reliability
Supported Parameters

This model supports the following parameters:

Top P Seed Temperature Stop Reasoning Max Tokens Include Reasoning Frequency Penalty Presence Penalty
Features

This model supports the following features:

Reasoning
Performance Summary

Baidu's ERNIE 4.5 21B A3B Thinking model, created in October 2025, demonstrates exceptional speed, consistently ranking among the fastest models across all 8 benchmarks. It also offers highly competitive pricing, placing it among the most cost-effective options. The model exhibits strong reliability with an 87% success rate, indicating consistent performance with few technical issues. In terms of performance across benchmarks, ERNIE 4.5 21B A3B Thinking shows particular strength in General Knowledge (97.0% accuracy) and Ethics (98.0% accuracy), suggesting a robust understanding of factual information and ethical principles. Its Reasoning capabilities are also commendable at 86.0% accuracy. However, the model struggles significantly with Instruction Following (0.0% accuracy) and Coding (17.3% accuracy), indicating substantial weaknesses in these areas. Its ability to acknowledge uncertainty in the Hallucinations test is moderate at 80.0%, while Mathematics performance is average at 68.0%. The model's description highlights its refinement for reasoning depth and quality, which is partially reflected in its strong Reasoning score, but its poor performance in Instruction Following and Coding suggests limitations in applying this reasoning to complex, multi-step tasks.

Model Pricing

Current Pricing

Feature Price (per 1M tokens)
Prompt $0.07
Completion $0.28

Price History

Available Endpoints
Provider Endpoint Name Context Length Pricing (Input) Pricing (Output)
Novita
Novita | baidu/ernie-4.5-21b-a3b-thinking 131K $0.07 / 1M tokens $0.28 / 1M tokens
Novita
Novita | baidu/ernie-4.5-21b-a3b-thinking 131K $0.07 / 1M tokens $0.28 / 1M tokens
Benchmark Results
Benchmark Category Reasoning Strategy Free Executions Accuracy Cost Duration
Other Models by baidu