ERNIE-4.5-21B-A3B-Thinking
ERNIE-4.5-21B-A3B-Thinking is a text-based Mixture of Experts (MoE) post-training model featuring 21B total parameters with 3B active parameters per token.
text → text · made by Baidu
$0.07→$0.28per Mtok in / out
Novita AI
Sold by 2 ways
| Seller | Lane | Rate |
|---|---|---|
| Novita AI 0 ways | $0.07per Mtok in$0.28per Mtok out | |
| Novita AI aggregatornot on the seller's list today | standard | $0.07per Mtok in$0.28per Mtok outapi.novita.ai · read 2026-08-25 |
| PPIO 0 ways | $0.5per Mtok in$2per Mtok out | |
| PPIO aggregatornot on the seller's list today | standard | $0.5per Mtok in$2per Mtok outapi.ppinfra.com · read 2026-09-10 |
About
ERNIE-4.5-21B-A3B-Thinking — a text model from Baidu.
It takes text and returns text, with a context window of 131,072 tokens. It was published in March 2025, trained on material up to March 2025. Its sellers say it can reason step by step. The catalogue files it under reasoning.
Every current figure
- Maker
- Baidu
- Register
- model
- Takes
- text
- Returns
- text
- Context
- 131,072 tokens
- Longest answer
- 65,536 tokens
- Published
- March 2025
- Knowledge to
- March 2025
- Parameters
- 21 billion · read from its own name, the total of a mixture
- Licence
- not read
Known as 1 name
baidu/ernie-4.5-21B-a3b-thinking