A sophisticated text-based Mixture-of-Experts (MoE) model featuring 21B total parameters with 3B activated per token, delivering exceptional multimodal understanding and generation through heterogeneous MoE structures and modality-isolated routing. Supporting an extensive 131K token context length, the model achieves efficient inference via multi-expert parallel collaboration and quantization, while advanced post-training techniques including SFT, DPO, and UPO ensure optimized performance across diverse applications with specialized routing and balancing losses for superior task handling.
| 信号 | 强度 | 权重 | 影响 |
|---|---|---|---|
| Recencyjust now | 92 | 15% | +13.8 |
| Context Windowjust now | 81 | 15% | +12.1 |
| Capabilitiesjust now | 33 | 30% | +10.0 |
| Output Capacityjust now | 65 | 15% | +9.8 |
| Pricingjust now | 0 | 25% | +0.1 |
社区和从业者反馈在基准测试和价格之上增加了真实世界的信号。
Share your experience with ERNIE 4.5 21B A3B and help the community make better decisions.
成本估算器
每月比类别平均节省$41.21