HunyuanOCR — Hardware Requirements & GPU Compatibility
Vision
HunyuanOCR is Tencent's 1.1-billion-parameter vision-language model built for OCR and document understanding, unifying document parsing, text spotting, information extraction, and text-image translation in one model rather than a general chat assistant. The current checkpoint, HunyuanOCR-1.5, adds DFlash speculative decoding, where a lightweight draft model proposes tokens that the main model verifies in one pass, cutting latency on long structured outputs like tables and formulas. It also ships GGUF weights for llama.cpp, running on CPUs and laptop or consumer GPUs, not just server-grade vLLM. At just over 1 billion parameters, it is light enough for a single modest consumer GPU. Context length is 131,072 tokens; pretraining supports image resolutions up to 4K. It is released under the Tencent Hunyuan Community License, a custom license permitting commercial use outside the EU, UK, and South Korea and below 100 million monthly active users. It was published in November 2025.
Runs great on a 24 GB GPU
Runs from 8 GB (Q4_K_M · 128K)Best on 24 GB (Q4_K_M · ~612 tok/s · 128K)Macs from 8 GB (Q4_K_M · ~67 tok/s)
Other
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run HunyuanOCR
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
ollama run hf.co/mradermacher/HunyuanOCR-GGUF:Q4_K_MQ4_K_M fits fully on the RTX 4090 · 22.2 GB free · 16K context
Download: HunyuanOCR.Q4_K_M.gguf0.35 GBmradermacher
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on RTX 3080 Ti (12 GB) (10 GB headroom) · Q4_K_M
- Generation speed
- ~554 tok/s
- generation speed
- Cost per 1M output tokens
- $0.061
- per 1M output tokens
How Much VRAM Does HunyuanOCR Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 0.9 GB | 2.4 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 128K context · ~745 tok/s | |
| 0.9 GB | 2.4 GB | 3-bit small — 3-bit small quantization | Fits: 128K context · ~736 tok/s | |
| 0.9 GB | 2.5 GB | 3-bit medium — 3-bit medium quantization | Fits: 128K context · ~690 tok/s | |
| 1.0 GB | 2.5 GB | 4-bit · legacy — 4-bit legacy quantization | Fits: 128K context · ~683 tok/s | |
| 1.1 GB | 2.6 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 128K context · ~612 tok/s | |
| 1.2 GB | 2.7 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 128K context · ~546 tok/s | |
| 1.3 GB | 2.8 GB | 6-bit, very good — 6-bit quantization, very good quality | Fits: 128K context · ~496 tok/s | |
| 1.5 GB | 3 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Fits: 128K context · ~431 tok/s |
Selected Q4_K_M
4-bit medium quantization1.8GBat 16K context
- Weights
- 0.7 GB
- KV cache
- 0.8 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 22.2 GB free · ~612 tok/s
≥50 tok/s needs ≥82 GB/s · RTX 5090 ~1089, RTX 3090 Ti ~612
Which Hardware Can Run HunyuanOCR?
Q4_K_M · 1.1 GBHunyuanOCR (Q4_K_M) requires 1.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 2+ GB is recommended. Using the full 131K context window can add up to 6.3 GB, bringing total usage to 7.4 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run HunyuanOCR, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB).
≤ 11 GB
≤ 11 GB tier →fits · 6.2 GB free · up to 128K ctx
16 GPUs · 1 Mac class · 2 device families
RTX 3080~462Simulate output on RTX 3080RTX 2080 Ti~374Simulate output on RTX 2080 TiRTX 3070 Ti~370Simulate output on RTX 3070 TiM3*8 GB~67Simulate output on iMac 24" M3 (8 GB)Orin Nano~62Simulate output on NVIDIA Jetson Orin Nano 8GB (Super)iPhone~45Simulate output on iPhone 1720 GB
fits · 18.2 GB free · up to 128K ctx
1 GPU
RX 7900 XT~449Simulate output on RX 7900 XT32–40 GB
fits · 30.2 GB free · up to 128K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~1089Simulate output on RTX 5090M5 Max36 GB~301Simulate output on Mac Studio M5 Max (36 GB)M532 GB~101Simulate output on MacBook Air 13" M5 (32 GB)Orin~124Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~82Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at Q4_K_M
Alternatives to HunyuanOCR
Hunyuan family · 6 models →Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| MiniCPM5 1BOpenBMB | — | 1.3 GB | ~662 | same size |
| MinerU2.5 Pro 2605 1.2Bopendatalab | — | 1.1 GB | ~642 | same size |
| MinerU2.5 Pro 2604 1.2Bopendatalab | — | 1.1 GB | ~642 | same size |
| InternVL3 5 1BOpenGVLab | — | 2.2 GB | ~936 | same size |
| TinyLlama 1.1B Chat v1.0TinyLlama | — | 1 GB | ~649 | same size |
| HRM Text 1Bsapientinc | — | 3.1 GB | ~228 | same size |
Specifications
- Publisher
- Tencent
- Family
- Hunyuan
- Parameters
- 1.1B
- Architecture
- HunYuanVLForConditionalGeneration
- Context Length
- 131,072 tokens
- Vocabulary Size
- 120,818
- Release Date
- 2025-11-18
- License
- Other
Frequently Asked Questions
How much VRAM does HunyuanOCR need?
HunyuanOCR requires 1.1 GB of VRAM at Q4_K_M, or 2.6 GB at BF16. Full 131K context adds up to 6.3 GB (7.4 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 1.1B × 4.8 bits ÷ 8 = 0.7 GB
KV Cache + Overhead ≈ 0.4 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 ≈ 6.7 GB (at full 131K context)
VRAM usage by quantization
What's the best quantization for HunyuanOCR?
For HunyuanOCR, Q4_K_M (1.1 GB) offers the best balance of quality and VRAM usage. Q5_K_S (1.2 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 0.7 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run HunyuanOCR on a Mac?
Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run HunyuanOCR. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 1.1 GB of usable unified memory (RAM minus macOS overhead).
Can I run HunyuanOCR locally?
Yes — HunyuanOCR can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is HunyuanOCR?
At Q4_K_M, HunyuanOCR can reach ~4486 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~612 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 1.07 × 0.65 = ~4860 tok/s
Estimated speed at Q4_K_M (1.1 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of HunyuanOCR?
At Q4_K_M, the download is about 0.67 GB. The full-precision BF16 version is 2.24 GB. The smallest option (IQ2_XXS) is 0.31 GB.
Which GPUs can run HunyuanOCR?
52 consumer GPUs can run HunyuanOCR at Q4_K_M. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 52 GPUs have plenty of headroom for comfortable inference.
Which devices can run HunyuanOCR?
163 devices with unified memory can run HunyuanOCR at Q4_K_M (1.1 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.