Can ERNIE 4.5 21B A3B PT run on the RTX A6000?
Yes. At 8K context the best fit is Q8_0, using about 22.6 GB of the 45.6 GB usable on the RTX A6000. Estimated speed ceiling: ~33 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | 128K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|---|
| 2-bit (Q2_K) • | 8.4 GB Fits | 8.6 GB Fits | 9.9 GB Fits | 15.1 GB Fits | 95 |
| 3-bit (Q3_K_M) • | 10.4 GB Fits | 10.6 GB Fits | 11.9 GB Fits | 17.2 GB Fits | 75 |
| 4-bit (Q4_K_M) • | 13.4 GB Fits | 13.6 GB Fits | 14.9 GB Fits | 20.2 GB Fits | 57 |
| 5-bit (Q5_K_M) • | 15.5 GB Fits | 15.7 GB Fits | 17.0 GB Fits | 22.3 GB Fits | 49 |
| 6-bit (Q6_K) • | 17.7 GB Fits | 17.9 GB Fits | 19.2 GB Fits | 24.4 GB Fits | 42 |
| 8-bit (Q8_0) • | 22.4 GB Fits | 22.6 GB Fits | 24.0 GB Fits | 29.2 GB Fits | 33 |
| 16-bit (F16/BF16) • | 41.5 GB Tight | 41.7 GB Tight | 43.0 GB Tight | 48.3 GB No | 18 |
• real GGUF file size from bartowski/baidu_ERNIE-4.5-21B-A3B-PT-GGUF. Speed is a bandwidth ceiling (768 GB/s), not a benchmark. Methodology.
What fits instead
- Cheapest listed GPU that runs ERNIE 4.5 21B A3B PT: Arc A770 16GB (16 GB, launch MSRP $349) at Q4_K_M, fully in VRAM at 8K context.
- Largest popular model that fits the RTX A6000: Qwen3.6 35B A3B (36B, MoE) at Q8_0, about 36.2 GB at 8K context.
"Runs" and "fits" mean the whole model in GPU memory at 8K context (4-bit or better for the model pick). Prices are US launch MSRPs, not today's street prices. "Popular" means the 60 most-downloaded models we track.
More
ERNIE 4.5 21B A3B PT on every GPU and Mac · Every model that runs on the RTX A6000 · Other models that fit 48 GB VRAM · Try other settings in the calculator
Guides: Best GPU for local LLMs by budget · Can I run gpt-oss, Llama 70B or Qwen3? · Mac vs PC for local LLMs
Data: Hugging Face Hub (baidu/ERNIE-4.5-21B-A3B-PT), NVIDIA specs. Model data last checked 2026-11-18; refreshed hourly.