ERNIEFrom the builder
ERNIE 5.1 Officially Released! Topping Multiple Leaderboards — A Model That Writes Better and Understands You More
Baidu released ERNIE 5.1, which builds on ERNIE 5.0's pre-training but cuts total parameters to about a third and active parameters to about half, with pre-training compute that Baidu puts at 6 percent of comparable models. Baidu says it beats DeepSeek-V4-Pro on the τ³-bench and SpreadsheetBench-Verified agent tests and scores 99.6 on AIME26 with tools, behind only Gemini 3.1 Pro. Technically, it extracts an optimal sub-network from ERNIE 5.0's elastic sub-model matrix and trains with a disaggregated, fully asynchronous reinforcement learning setup on PaddlePaddle.
ERNIE 5.1 is derived from ERNIE 5.0, extracting the optimal sub-network architecture from ERNIE 5.0’s multi-dimensional elastic sub-model matrix to effectively inherit the knowledge and capabilities encoded in ERNIE 5.0 while significantly reducing pre-training cost.
Saved . A shortened excerpt from the official blog; the original publisher retains authorship.