Can I run ERNIE 5.5 Flash Base locally?
ERNIE 5.5 Flash Base by baidu has 286B parameters (5.7B active per token, 512 experts, 10 used per token). At 4-bit with an 8K context it needs about 164 GB of GPU memory. It fits on 6 of 65 GPUs and Macs we track.
This model uses a newer compressed or sparse attention design. Our KV-cache numbers use the standard formula, so treat long-context figures as an upper bound.
Memory needed by quant and context length
GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx | 8K ctx | 32K ctx | 128K ctx |
|---|---|---|---|---|---|
| 2-bit (Q2_K) | rough | 101 GB | 101 GB | 103 GB | 108 GB |
| 3-bit (Q3_K_M) | usable | 131 GB | 132 GB | 133 GB | 138 GB |
| 4-bit (Q4_K_M) | good, most popular | 164 GB | 164 GB | 166 GB | 171 GB |
| 5-bit (Q5_K_M) | very good | 191 GB | 191 GB | 193 GB | 198 GB |
| 6-bit (Q6_K) | near lossless | 220 GB | 220 GB | 221 GB | 227 GB |
| 8-bit (Q8_0) | effectively lossless | 284 GB | 284 GB | 286 GB | 291 GB |
| 16-bit (F16/BF16) | original weights | 534 GB | 534 GB | 536 GB | 541 GB |
Which GPUs and Macs run ERNIE 5.5 Flash Base?
Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).
Questions
How much VRAM does ERNIE 5.5 Flash Base need?
About 164 GB at 4-bit (Q4_K_M) with an 8K context: 163 GB of weights (estimated), 0.5 GB of KV cache and 0.8 GB of runtime overhead. At 8-bit it needs about 284 GB.
What is the smallest GPU that runs ERNIE 5.5 Flash Base?
No single consumer GPU in our table fits it at Q4_K_M with 8K context; the smallest option that does is the Mac M3 Ultra 256GB (256 GB).
Can I run ERNIE 5.5 Flash Base on a Mac?
Yes. At Q4_K_M with 8K context it fits on a Mac M3 Ultra 256GB or larger, using the default macOS GPU memory limit.
Similar models
Same family or a similar parameter count, with the VRAM each needs at Q4_K_M and 8K context.
ERNIE 4.5 21B A3B PT
22B · 13.5 GB at 4-bitERNIE 4.5 21B A3B Thinking
22B · 13.5 GB at 4-bitQianfan OCR
4.7B · 4.4 GB at 4-bitDeepSeek V4 Flash
291B · 167 GB at 4-bitHy3
299B · 173 GB at 4-bitHy3 preview
299B · 173 GB at 4-bitModel facts
| Hugging Face repo | baidu/ERNIE-5.5-Flash-Base |
| Parameters | 286,245,697,882 (286B) · ~5.7B active (computed from config), low confidence, speed not shown |
| Layers / hidden size | 43 / 4096 |
| Attention | MLA; 64 heads, 1 KV heads, head dim 512 |
| KV cache per token (FP16) | 59.1 KiB |
| Max context | 1,048,576 tokens |
| License | other |
| Released on HF | Oct 21, 2026 |
| Downloads (30d) / likes | 4,145 / 353 |
Sources: Hugging Face model API · config.json (baidu/ERNIE-5.5-Flash-Base) · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.