Hy3 Preview — Hardware Requirements & GPU Compatibility
ChatFunctions
Hy3 Preview is a 298.8B-parameter open language model from Tencent in the Hunyuan 3 family. It supports a context window of up to 262,144 tokens. At Q4_K_M it needs about 179.91 GB of VRAM — see which GPUs and Macs can run it below.
Hy3 Preview runs on a 192 GB+ Mac or a datacenter GPU
Best local pick → Mac Pro M2 Ultra (192 GB) · Q4_K_M · 182.3 GB · ~26 tok/s
Macs from 192 GB (Q4_K_M · ~26 tok/s)
Other
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Hy3 Preview
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
vllm serve tencent/Hy3-preview --max-model-len 16384 --tensor-parallel-size 8vLLM repo with its own quantisation
Download: Weights on Hugging Face
Other ways
each opens step-by-stepRun in cloud
Fits on B300 (288 GB) (105 GB headroom) · Q4_K_M
- Generation speed
- ~147 tok/s
- generation speed
- Cost per 1M output tokens
- $13.16
- per 1M output tokens
How Much VRAM Does Hy3 Preview Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | Runs on | |
|---|---|---|---|---|
| 127.6 GB | 132.7 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Runs on H200 NVL 141 GB | |
| 146.3 GB | 151.3 GB | 3-bit medium — 3-bit medium quantization | Runs on B200 192 GB | |
| 179.9 GB | 184.9 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Runs on B200 192 GB | |
| 213.5 GB | 218.6 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Runs on Instinct MI325X 256 GB | |
| 247.1 GB | 252.2 GB | 6-bit, very good — 6-bit quantization, very good quality | Runs on Instinct MI325X 256 GB | |
| 299.4 GB | 304.5 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Runs on 5× H100 SXM 80 GB | |
| 598.2 GB | 603.2 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Runs on 5× H200 SXM 141 GB |
Selected Q4_K_M
4-bit medium quantization182.3GBat 16K context
- Weights
- 179.3 GB
- KV cache
- 2.7 GB
- Overhead
- 0.3 GB
Server-class: ~182.3 GB at 16K context · one B200 192 GB
Which Hardware Can Run Hy3 Preview?
Q4_K_M · 179.9 GBHy3 Preview (Q4_K_M) requires 179.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 234+ GB is recommended. Using the full 262K context window can add up to 42.6 GB, bringing total usage to 222.5 GB. No consumer GPU has enough memory. 4 devices with unified memory can run Hy3 Preview, including Mac Studio (M3 Ultra, 512GB), Mac Studio (M3 Ultra, 256GB).
192+ GB
fits · 9.7 GB free · up to 64K ctx
6 GPUs · 1 Mac class · 2 device families
B200~147Simulate output on B200B300~147Simulate output on B300Instinct MI350X~54Simulate output on Instinct MI350XM3 Ultra256–512 GB~26Simulate output on Mac Studio (M3 Ultra, 256GB)8x H100 SXM5~195Simulate output on NVIDIA DGX H1008x A100 SXM4~179Simulate output on NVIDIA DGX A100 640GB
By VRAM tier · typical GPUs and Macs at Q4_K_M
Consumer GPUs (up to 32 GB) and Macs up to 128 GB: too small for this model (needs ≥ 130 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →
How good is Hy3 Preview?
llmrun Score blends public benchmark boards into one number; 100 = the average open model.
Not enough data yet for an llmrun Score.
A Score needs results on enough independent boards. How the Composite Score is built
Benchmarks
Here are all 2 boards that have results for this model, sorted by their weight in the llmrun Score; badges show who ran the test.
- LMArena Text
LMArena · Oct 2026Arena
Arena1409.1
#44/187 open
- LMArena WebDev
LMArena · Oct 2026Arena
Arena1356.9
#37/52 open
Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| GLM 5.3 FlashZ.ai | 124.2 | 205.2 GB | —needs > 24 GB | same size |
| IQuest Q1IQuestLab | — | 195.4 GB | —needs > 24 GB | same size |
| MiMo V2.6 Flash MOPDXiaomi | — | 186.5 GB | —needs > 24 GB | same size |
| DeepSeek V4 Flash Vision ExpDeepSeek | — | 183.3 GB | —needs > 24 GB | same size |
| Step3StepFun | — | 219.1 GB | —needs > 24 GB | same size |
| Motif 3Motif-Technologies | — | 632.8 GB | —needs > 24 GB | same size |
Specifications
- Publisher
- Tencent
- Family
- Hunyuan 3
- Parameters
- 298.8B
- Active parameters
- 21B per token
- Architecture
- HYV3ForCausalLM
- Context Length
- 262,144 tokens
- Vocabulary Size
- 120,832
- Release Date
- 2026-04-13
- License
- Other
Frequently Asked Questions
How much VRAM does Hy3 Preview need?
Hy3 Preview requires 179.9 GB of VRAM at Q4_K_M, or 598.2 GB at BF16. Full 262K context adds up to 42.6 GB (222.5 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 298.8B × 4.8 bits ÷ 8 = 179.3 GB
KV Cache + Overhead ≈ 0.6 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
KV Cache + Overhead ≈ 43.2 GB (at full 262K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run Hy3 Preview?
No — Hy3 Preview requires at least 127.6 GB at Q2_K, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
What's the best quantization for Hy3 Preview?
For Hy3 Preview, Q4_K_M (179.9 GB) offers the best balance of quality and VRAM usage. Q5_K_M (213.5 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 127.6 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Hy3 Preview on a Mac?
Yes — Mac Pro M2 Ultra (192 GB) and 3 other Macs can run Hy3 Preview. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 179.9 GB of usable unified memory (RAM minus macOS overhead).
Can I run Hy3 Preview locally?
Yes — Hy3 Preview can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Hy3 Preview?
At Q4_K_M, Hy3 Preview can reach ~54 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)
Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.
Example: NVIDIA B200 → 12.6 GB active ÷ (8000 × 0.65) = 2.42655 ms, plus 80 layers × 0.055 ms = 4.40000 ms, so 1000 ÷ 6.82655 ms = ~146 tok/s
Estimated speed at Q4_K_M (179.9 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Hy3 Preview?
At Q4_K_M, the download is about 179.27 GB. The full-precision BF16 version is 597.57 GB. The smallest option (Q2_K) is 126.98 GB.
Which GPUs can run Hy3 Preview?
No single consumer GPU has enough VRAM to run Hy3 Preview at Q4_K_M (179.9 GB). Multi-GPU or professional hardware is required.
Which devices can run Hy3 Preview?
4 devices with unified memory can run Hy3 Preview at Q4_K_M (179.9 GB), including Mac Pro M2 Ultra (192 GB), Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M2 Ultra (192 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.