ERNIE-5.0
ERNIE 5.0 is Baidu's 2.4-trillion-parameter natively omni-modal foundation model, mapping text, images, video and audio into a shared token space within a single autoregressive framework for unified multimodal understanding and generation.
text + image + audio + video → text · made by Baidu
Sold by 2 ways
| Seller | Lane | Rate |
|---|---|---|
| Baidu cloud | standard | $0.89per Mtok in$3.56per Mtok outcloud.baidu.com · read 2026-08-25 |
| Baidu cloud | 32k-128k | $1.48per Mtok in$5.94per Mtok outcloud.baidu.com · read 2026-08-25 |
About
ERNIE-5.0 — a text model from Baidu, sold by one company from $0.89 in and $3.56 out per million tokens.
It takes text, images, audio and video and returns text, with a context window of 128,000 tokens. It was published in January 2026, trained on material up to 2025. Its sellers say it can reason step by step and call a tool. The catalogue files it under extract. Only Baidu sells it, at $0.89 in and $3.56 out per million tokens. Beside the standard rate there is a separately routed lane.
Every current figure
- Maker
- Baidu
- Register
- model
- Takes
- text + image + audio + video
- Returns
- text
- Context
- 128,000 tokens
- Longest answer
- 64,000 tokens
- Published
- January 2026
- Knowledge to
- 2025
- Licence
- not read
- Sellers
- 1
- Price
- $0.89 in and $3.56 out per million tokens — Baidu