Can I run Hunyuan A13B Instruct locally?
Hunyuan A13B Instruct by tencent has 80B parameters (13B active per token, 64 experts, 88888888888888888888888888888888 used per token). At 4-bit with an 8K context it needs about 47.1 GB of GPU memory. It fits on 25 of 65 GPUs and Macs we track. See all models that fit in 64 GB of VRAM.
Memory needed by quant and context length
GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from unsloth/Hunyuan-A13B-Instruct-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx | 8K ctx | 32K ctx |
|---|---|---|---|---|
| 2-bit (Q2_K) • | rough | 28.5 GB | 29.0 GB | 32.0 GB |
| 3-bit (Q3_K_M) • | usable | 37.1 GB | 37.6 GB | 40.6 GB |
| 4-bit (Q4_K_M) • | good, most popular | 46.6 GB | 47.1 GB | 50.1 GB |
| 5-bit (Q5_K_M) • | very good | 54.3 GB | 54.8 GB | 57.8 GB |
| 6-bit (Q6_K) | near lossless | 62.5 GB | 63.0 GB | 66.0 GB |
| 8-bit (Q8_0) • | effectively lossless | 80.7 GB | 81.2 GB | 84.2 GB |
| 16-bit (F16/BF16) • | original weights | 151 GB | 151 GB | 154 GB |
Which GPUs and Macs run Hunyuan A13B Instruct?
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 Hunyuan A13B Instruct need?
About 47.1 GB at 4-bit (Q4_K_M) with an 8K context: 45.4 GB of weights (real GGUF file size), 1.0 GB of KV cache and 0.6 GB of runtime overhead. At 8-bit it needs about 81.2 GB.
What is the smallest GPU that runs Hunyuan A13B Instruct?
No single consumer GPU in our table fits it at Q4_K_M with 8K context; the smallest option that does is the H100 PCIe (80 GB).
Can I run Hunyuan A13B Instruct on a Mac?
Yes. At Q4_K_M with 8K context it fits on a Mac M2 Max 96GB 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.
Hy Memory 30B A3B
30B · 18.5 GB at 4-bitWeVisOne VL 4B
4.8B · 4.3 GB at 4-bitWeVisOne VL 2B
2.4B · 2.8 GB at 4-bitKimi Dev 72B
73B · 44.6 GB at 4-bitLlama 3.3 70B Instruct
71B · 43.3 GB at 4-bitDeepSeek R1 Distill Llama 70B
71B · 43.3 GB at 4-bitModel facts
| Hugging Face repo | tencent/Hunyuan-A13B-Instruct |
| Parameters | 80,393,183,232 (80B) · ~13B active (from the model name) |
| Layers / hidden size | 32 / 4096 |
| Attention | GQA; 32 heads, 8 KV heads, head dim 128 |
| KV cache per token (FP16) | 128.0 KiB |
| Max context | 32,768 tokens |
| License | other |
| Released on HF | Jun 25, 2025 |
| Downloads (30d) / likes | 74,022 / 795 |
Sources: Hugging Face model API · config.json (tencent/Hunyuan-A13B-Instruct) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.