Grug 12B — Hardware Requirements & GPU Compatibility
ChatReasoning
Grug 12B is a 12.0B-parameter open language model from kai-os. It supports a context window of up to 262,144 tokens. At BF16 it needs about 24.97 GB of VRAM — see which GPUs and Macs can run it below.
Doesn't fit a 24 GB GPU
Smallest consumer fit → Snapdragon X Elite Copilot+ PC · BF16 · 25.8 GB · ~4 tok/s
Macs from 32 GB (BF16 · ~11 tok/s)
Other
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Grug 12B
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
vllm serve kai-os/Grug-12B --max-model-len 16384vLLM repo with its own quantisation
Download: Weights on Hugging Face
Other ways
each opens step-by-stepRun in cloud
Fits on RTX A6000 (48 GB) (22 GB headroom) · BF16
- Generation speed
- ~20 tok/s
- generation speed
- Cost per 1M output tokens
- $4.58
- per 1M output tokens
How Much VRAM Does Grug 12B Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 25.0 GB | 27 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Does not fit: needs 25.8 GB |
Selected BF16
Brain floating point 1625.8GBat 16K context
- Weights
- 23.9 GB
- KV cache
- 1.6 GB
- Overhead
- 0.3 GB
Doesn't fit RTX 4090 (reference) · needs 1.8 GB more
≥30 tok/s needs ≥1.15 TB/s · RTX 5090 ~47
Which Hardware Can Run Grug 12B?
BF16 · 25.0 GBGrug 12B (BF16) requires 25.0 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 33+ GB is recommended. Using the full 262K context window can add up to 19.7 GB, bringing total usage to 44.6 GB. 1 GPU can run it, including NVIDIA GeForce RTX 5090. 98 devices with unified memory can run Grug 12B, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M5 Max (36 GB).
32–40 GB
fits · 6.2 GB free · up to 64K ctx
1 GPU · 3 Mac classes · 2 device families
RTX 5090~47Simulate output on RTX 5090M5 Max36 GB~13Simulate output on Mac Studio M5 Max (36 GB)M532 GB~4Simulate output on MacBook Air 13" M5 (32 GB)Orin~5Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~4Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at BF16
Similar models
5| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Gemma 3 12B ITGoogle | 74.2 | 10.9 GB | ~82 | same size |
| Gemma 4 12BGoogle | — | 9 GB | ~80 | same size |
| Mistral Nemo Base 2407Mistral AI | — | 11 GB | ~81 | same size |
| Falcon 11BTII UAE | — | 8.5 GB | ~89 | same size |
| NVIDIA Nemotron Nano 12B v2NVIDIA | — | 11 GB | ~81 | same size |
Specifications
- Publisher
- kai-os
- Parameters
- 12.0B
- Architecture
- Gemma4UnifiedForConditionalGeneration
- Context Length
- 262,144 tokens
- Vocabulary Size
- 262,144
- Release Date
- 2026-07-02
- License
- Other
Frequently Asked Questions
How much VRAM does Grug 12B need?
Grug 12B requires 25.0 GB of VRAM at BF16. Full 262K context adds up to 19.7 GB (44.6 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 12.0B × 16 bits ÷ 8 = 23.9 GB
KV Cache + Overhead ≈ 1.1 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 ≈ 20.8 GB (at full 262K context)
VRAM usage by quantization
Can I run Grug 12B on a Mac?
Yes — Mac Studio M1 Max (32 GB) and 84 other Macs can run Grug 12B. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 25.0 GB of usable unified memory (RAM minus macOS overhead).
Can I run Grug 12B locally?
Yes — Grug 12B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Grug 12B?
At BF16, Grug 12B can reach ~192 tok/s on AMD Instinct MI350X. 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 ÷ 25.0 × 0.65 = ~208 tok/s
Estimated speed at BF16 (25.0 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Grug 12B?
At BF16, the download is about 23.92 GB.
Which GPUs can run Grug 12B?
1 consumer GPU can run Grug 12B at BF16. Top options include NVIDIA GeForce RTX 5090.
Which devices can run Grug 12B?
98 devices with unified memory can run Grug 12B at BF16 (25.0 GB), including ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB), GMKtec EVO-X2 (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.