by
A powerful multimodal Mixture-of-Experts chat model featuring 28B total parameters with 3B activated per token, delivering exceptional text and vision understanding through its innovative heterogeneous MoE structure with modality-isolated routing. Built with scaling-efficient infrastructure for high-throughput training and inference, the model leverages advanced post-training techniques including SFT, DPO, and UPO for optimized performance, while supporting an impressive 131K context length and RLVR alignment for superior cross-modal reasoning and generation capabilities.
| Signal | Strength | Weight | Impact |
|---|---|---|---|
| Capabilitiesjust now | 67 | 30% | +20.0 |
| Recencyjust now | 92 | 15% | +13.9 |
| Context Windowjust now | 71 | 15% | +10.7 |
| Output Capacityjust now | 65 | 15% | +9.8 |
| Pricingjust now | 1 | 25% | +0.1 |
Community and practitioner feedback adds real-world signal on top of benchmarks and pricing.
Share your experience with ERNIE 4.5 VL 28B A3B and help the community make better decisions.
Cost Estimator
You save $39.31/month vs category average