ERNIE-4.5-VL-424B-A47B is a multimodal Mixture-of-Experts (MoE) model from Baidu’s ERNIE 4.5 series, featuring 424B total parameters with 47B active per token. It is trained jointly on text and image data using a heterogeneous MoE architecture and modality-isolated routing to enable high-fidelity cross-modal reasoning, image understanding, and long-context generation (up to 131k tokens). Fine-tuned with techniques like SFT, DPO, UPO, and RLVR, this model supports both “thinking” and non-thinking inference modes. Designed for vision-language tasks in English and Chinese, it is optimized for efficient scaling and can operate under 4-bit/8-bit quantization.
| 信号 | 强度 | 权重 | 影响 |
|---|---|---|---|
| Capabilitiesjust now | 50 | 30% | +15.0 |
| Recencyjust now | 84 | 15% | +12.6 |
| Context Windowjust now | 81 | 15% | +12.1 |
| Output Capacityjust now | 70 | 15% | +10.5 |
| Pricingjust now | 1 | 25% | +0.3 |
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 VL 424B A47B and help the community make better decisions.
成本估算器
每月比类别平均节省$35.85