// Open-model VRAM table · family
Baidu ERNIE: local deployment & VRAM
11 open-weight models from Baidu (16 official repos incl. quantised builds). At 4-bit they range from 0.9 GB (ERNIE-4.5-0.3B-PT) to 285 GB (ERNIE-4.5-VL-424B-A47B-PT).
official orgs: baidulatest 2026-04-07memory = estimate
01Text LLM (5)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| ERNIE-4.5-21B-A3B-Thinking | 21.8B / 3B | 128K | 49.0 GB | 15.5 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-09-08 |
| ERNIE-4.5-0.3B-PT | 361M | 128K | 1.4 GB | 0.9 GB | 8GB GPU | Mac 16GB | Commercial OK | 2025-06-28 |
| ERNIE-4.5-21B-A3B-PT | 21.9B / 3B | 128K | 49.3 GB | 15.6 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-06-28 |
| ERNIE-4.5-300B-A47B-PT | 300.5B / 47B | 128K | 663 GB | 203 GB | 4 × RTX PRO 6000 (384GB) | Mac 512GB | Commercial OK | 2025-06-28 |
| ERNIE-4.5-300B-A47B-Paddle | 300.5B / 47B | 128K | 663 GB | 203 GB | 4 × RTX PRO 6000 (384GB) | Mac 512GB | Commercial OK | 2025-06-28 |
02Vision-language (4)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| ERNIE-4.5-VL-28B-A3B-Thinking ↗ | 29.7B / 3B | 128K | 66.2 GB | 20.8 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-11-07 |
| ERNIE-4.5-VL-28B-A3B-PT | 29.4B / 3B | 128K | 65.6 GB | 20.6 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2025-06-28 |
| ERNIE-4.5-VL-424B-A47B-PT | 423.5B / 47B | 128K | 934 GB | 285 GB | 4 × RTX PRO 6000 (384GB) | Mac 512GB | Commercial OK | 2025-06-28 |
| ERNIE-4.5-VL-424B-A47B-Paddle | 423.5B / 47B | 128K | 934 GB | 285 GB | 4 × RTX PRO 6000 (384GB) | Mac 512GB | Commercial OK | 2025-06-28 |
03Image generation (2)
| Model | Params | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|
| ERNIE-Image ↗ | 8.03B | 18.2 GB | 5.9 GB | 8GB GPU | Mac 16GB | Commercial OK | 2026-04-07 |
| ERNIE-Image-Turbo ↗ | 8.03B | 18.2 GB | 5.9 GB | 8GB GPU | Mac 16GB | Commercial OK | 2026-04-02 |
04Other families
Qwen · IBM Granite · NVIDIA Nemotron · Gemma · Mistral / Magistral / Devstral · GLM / CogVideoX (Z.ai) · DeepSeek · Ai2 OLMo / Molmo · Ant Ling / Ring / Ming · Liquid LFM · MiniCPM / VoxCPM · Tencent Hunyuan / Hy · Phi · Llama · Meituan LongCat · Jina embeddings & rerankers · Wan · Kimi · Cohere Command / Aya · Xiaomi MiMo · FLUX · Microsoft VibeVoice / Fara / Harrier · InternLM / Intern-S · StepFun Step · MiniMax · NVIDIA Parakeet / Canary speech · Stable Diffusion / Stable Audio · Yi (01.AI) · ByteDance Seed · SmolLM / SmolVLM · BGE embeddings & rerankers · LTX Video · FunAudioLLM speech · gpt-oss · poolside Laguna · Whisper · Baichuan · Thinking Machines Inkling · xAI Grok · rednote dots
Totals include 10% + 0.5GB overhead and one 8K-token KV cache for text models. Estimates only. Method →