// Open-model VRAM table · family
Tencent Hunyuan / Hy: local deployment & VRAM
19 open-weight models from Tencent (56 official repos incl. quantised builds). At 4-bit they range from 1.7 GB (Hunyuan-0.5B-Instruct) to 515 GB (Hy4-preview).
official orgs: tencentlatest 2026-10-15memory = estimate
01Text LLM (10)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| Hy-Memory-30B-A3B | 30.1B / 3B | 256K | 67.5 GB | 21.4 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2026-10-15 |
| Hy4-preview | 770B / 49B | 1M | 1.70 TB | 515 GB | 8 × H100 (640GB) | — | Commercial OK | 2026-08-27 |
| Hy3 | 295B / 21B | 256K | 652 GB | 200 GB | 4 × RTX PRO 6000 (384GB) | Mac 512GB | Commercial OK | 2026-07-02 |
| Hy3-preview | 295B / 21B | 256K | 652 GB | 200 GB | 4 × RTX PRO 6000 (384GB) | Mac 512GB | Custom licence – read it | 2026-04-13 |
| Hunyuan3D-Part | — | — | — | — | — | — | Licence not stated | 2025-09-23 |
| Hunyuan-0.5B-Instruct ↗ | 539M | 256K | 2.5 GB | 1.7 GB | 8GB GPU | Mac 16GB | Licence not stated | 2025-07-30 |
| Hunyuan-1.8B-Instruct | 1.79B | 256K | 5.0 GB | 2.2 GB | 8GB GPU | Mac 16GB | Licence not stated | 2025-07-30 |
| Hunyuan-4B-Instruct ↗ | 4.22B | 256K | 11.0 GB | 4.5 GB | 8GB GPU | Mac 16GB | Licence not stated | 2025-07-30 |
| Hunyuan-7B-Instruct | 7.5B | 32K | 18.1 GB | 6.6 GB | 8GB GPU | Mac 16GB | Licence not stated | 2025-07-30 |
| Hunyuan-A13B-Instruct | 80.4B / 13B | 32K | 178 GB | 55.2 GB | 2 × RTX 5090 (64GB) | Mac 128GB | Commercial with conditions | 2025-06-25 |
02Vision-language (1)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| HunyuanOCR | 1.12B | 128K | 3.8 GB | 2.1 GB | 8GB GPU | Mac 16GB | Commercial with conditions | 2025-11-18 |
03Image generation (1)
| Model | Params | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|
| HunyuanImage-3.0-Instruct | 80B / 13B | 177 GB | 53.9 GB | 2 × RTX 5090 (64GB) | Mac 128GB | Custom licence – read it | 2025-09-25 |
04Video generation (2)
| Model | Params | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|
| HunyuanVideo-1.5 ↗ | — | — | — | — | — | Commercial with conditions | 2025-11-18 |
| HunyuanVideo ↗ | — | — | — | — | — | Commercial with conditions | 2024-12-01 |
05Translation (5)
| Model | Params | Context | BF16 | 4-bit | Smallest GPU (4-bit) | Mac (4-bit) | Licence | Released |
|---|---|---|---|---|---|---|---|---|
| Hy-MT2-1.8B | 2.04B | 256K | 5.5 GB | 2.4 GB | 8GB GPU | Mac 16GB | Commercial OK | 2026-05-11 |
| Hy-MT2-7B | 8.03B | 256K | 19.2 GB | 6.9 GB | 8GB GPU | Mac 16GB | Commercial OK | 2026-05-11 |
| Hy-MT2-30B-A3B | 30.1B / 3B | 256K | 67.5 GB | 21.4 GB | RTX 4090 24GB | Mac 36GB | Commercial OK | 2026-05-11 |
| HY-MT1.5-1.8B | 2.04B | 256K | 5.5 GB | 2.4 GB | 8GB GPU | Mac 16GB | Licence not stated | 2025-12-25 |
| Hunyuan-MT-7B ↗ | 8.03B | 32K | 19.2 GB | 6.9 GB | 8GB GPU | Mac 16GB | Licence not stated | 2025-08-28 |
06Other 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 · 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 · Baidu ERNIE · 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 →