Architecture
MoE
Released
2025-06-30
Layers
28
KV Heads
4
Head Dim
128
Family
ernie
Quantization Options
Context length:
| Quant | Bits | VRAM @ 16K | Quality |
|---|---|---|---|
| Q3_K_M | 4 | 12.1 GB 11.5 + 0.7 KV | low |
| Q3_K_L | 4.3 | 12.9 GB 12.3 + 0.7 KV | moderate |
| IQ4_XS | 4.46 | 13.4 GB 12.7 + 0.7 KV | moderate |
| Q4_K_S | 4.67 | 14.0 GB 13.3 + 0.7 KV | moderate |
| Q4_K_M | 4.89 | 14.6 GB 13.9 + 0.7 KV | good |
| Q5_K_S | 5.57 | 16.4 GB 15.8 + 0.7 KV | good |
| Q5_K_M | 5.7 | 16.8 GB 16.1 + 0.7 KV | good |
| Q6_K | 6.56 | 19.1 GB 18.5 + 0.7 KV | excellent |
| Q8_0 | 8.5 | 24.5 GB 23.8 + 0.7 KV | lossless |
| FP16 | 16 | 45.0 GB 44.4 + 0.7 KV | lossless |
Select your GPU above to see speed estimates and compatibility for each quantization.
Ready to run this?Rent by the hour
Rent a GPU and run ERNIE 4.5 21B A3B now
Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.
Community Ratings
Loading ratings...
Benchmarks (3)
Run this model
>_
Auto-setup with fitmyllm CLI
Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.
Auto-detect GPULive tok/s in chatSpeed benchmarks9 inference engines
GPUs that can run this model
At Q4_K_M quantization. Sorted by minimum VRAM.
NVIDIA RTX 5080
16 GB VRAM • 960 GB/s
NVIDIA
$999
NVIDIA RTX 5070 Ti
16 GB VRAM • 896 GB/s
NVIDIA
$749
NVIDIA RTX 4080 SUPER
16 GB VRAM • 736 GB/s
NVIDIA
$999
NVIDIA RTX 4080
16 GB VRAM • 717 GB/s
NVIDIA
$1199
NVIDIA RTX 4070 Ti SUPER
16 GB VRAM • 672 GB/s
NVIDIA
$799
NVIDIA RTX 4060 Ti 16GB
16 GB VRAM • 288 GB/s
NVIDIA
$499
AMD RX 7900 GRE
16 GB VRAM • 576 GB/s
AMD
$549
AMD RX 7800 XT
16 GB VRAM • 624 GB/s
AMD
$499
AMD RX 7600 XT
16 GB VRAM • 288 GB/s
AMD
$329
AMD RX 6950 XT
16 GB VRAM • 576 GB/s
AMD
$449
AMD RX 6900 XT
16 GB VRAM • 512 GB/s
AMD
$469
AMD RX 6800 XT
16 GB VRAM • 512 GB/s
AMD
$599
AMD RX 6800
16 GB VRAM • 512 GB/s
AMD
$599
Intel Arc A770 16GB
16 GB VRAM • 560 GB/s
INTEL
$349
Apple M1 Pro (16GB)
16 GB VRAM • 200 GB/s
APPLE
$999
Apple M2 Pro (16GB)
16 GB VRAM • 200 GB/s
APPLE
$1299
Apple M4 (16GB)
16 GB VRAM • 120 GB/s
APPLE
$499
NVIDIA Tesla T4 16GB
16 GB VRAM • 320 GB/s
NVIDIA
$800
NVIDIA V100 PCIe 16GB
16 GB VRAM • 900 GB/s
NVIDIA
$2000
AMD RX 9070 XT
16 GB VRAM • 640 GB/s
AMD
$599
AMD RX 9070
16 GB VRAM • 672 GB/s
AMD
$549
Apple M1 (16GB)
16 GB VRAM • 68.25 GB/s
APPLE
$699
Apple M2 (16GB)
16 GB VRAM • 100 GB/s
APPLE
$799
Apple M3 (16GB)
16 GB VRAM • 100 GB/s
APPLE
$799
NVIDIA RTX 5060 Ti 16GB
16 GB VRAM • 448 GB/s
NVIDIA
$429
NVIDIA Tesla P100 DGXS
16 GB VRAM • 732 GB/s
NVIDIA
NVIDIA Tesla P100 PCIe 16 GB
16 GB VRAM • 732 GB/s
NVIDIA
NVIDIA Tesla P100 SXM2
16 GB VRAM • 732 GB/s
NVIDIA
NVIDIA Tesla V100 PCIe 16 GB
16 GB VRAM • 897 GB/s
NVIDIA
NVIDIA Tesla V100 SXM2 16 GB
16 GB VRAM • 1130 GB/s
NVIDIA
Find the best GPU for ERNIE 4.5 21B A3B
Build Hardware for ERNIE 4.5 21B A3BModel Card
View on HuggingFaceBaidu ERNIE 4.5 — MoE model with 3B active. Fast Chinese-English generation.