Model profile
Ernie 4.5 300B A47B
Baidudeveloper
2025-06-28release date
#127 / 267overall rank
5eval lineages
Evidence summary
Ernie 4.5 300B A47B has an estimated overall rank of #127; its 90% source-sensitivity interval is #26–#208. Its behavior-only rank is #132; company governance moves the combined estimate to #127. Published evidence spans 5 evals and 6 of 7 behavior components. Its strongest relative result is FlagEval Safety and Values (a5_qualified_rate, #1 of 18); its weakest is BullshitBench v2 (clear_pushback_rate, #100 of 105).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗ernie-4-5-300b-a47b
- Hugging Face ↗baidu/ERNIE-4.5-300B-A47B-PT
- OpenRouter ↗baidu/ernie-4.5-300b-a47b
- Official model page ↗Exact model document · Reviewed official Hugging Face owner · official repository
- Release source ↗repository created at proxy
Published eval results
Rank is within that sub-eval. Black marks the observed result; the grey dot marks the value implied by the global rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Better | Source |
|---|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #90 / 311 | 0.671 | ↓ lower | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #100 / 105 | 0.04 | ↑ higher | Source ↗official | |
| Confabulationsconfabulation_rate | #28 / 52 | 25.25 | ↓ lower | Source ↗official | |
| FlagEval Safety and Valuesa1_qualified_rate | #2 / 18 | 85.04 | ↑ higher | Source ↗official | |
| FlagEval Safety and Valuesa2_qualified_rate | #11 / 18 | 78.79 | ↑ higher | Source ↗official | |
| FlagEval Safety and Valuesa3_qualified_rate | #13 / 18 | 85.37 | ↑ higher | Source ↗official | |
| FlagEval Safety and Valuesa4_qualified_rate | #3 / 18 | 90.04 | ↑ higher | Source ↗official | |
| FlagEval Safety and Valuesa5_qualified_rate | #1 / 18 | 77.55 | ↑ higher | Source ↗official | |
| UAVBench safety-critical decision recognitionethical_safety_critical_accuracy | #16 / 27 | 0.68 | ↑ higher | Source ↗official |