GameScaling 27B S3 needs about 19.2 GB of memory at Q4_K_M (a 16.5 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 11 quants, from 10.7 GB (Q2_K) to 28.6 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 8.7 GB.
GameScaling 27B S3 – Qwen3.5 vision‑language model for game action prediction
GameScaling 27B S3 is a Qwen3.5‑27B model continued pre‑trained on 1,600 h of gameplay. It predicts the next 200 ms of keyboard and mouse actions from a game frame and short history. The model is released under apache‑2.0 and supports reasoning, tools, and vision.
Good at
- Game action prediction
- Vision‑language reasoning
- Tool usage
Limits
- No training code provided
- Requires large VRAM for full precision
- No vision support in non‑vision quants
Best for
- Game AI prototyping
- Real‑time action generation
- Embedded game assistants
Who it’s for. Developers building game‑AI agents or researchers studying embodied agents. The model is suitable for those who need a vision‑language model that can output precise keyboard and mouse actions based on game frames.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 10.7 GB | 13.4 GB | Too big | Get file → | |
| Q3_K_S | 12.1 GB | 14.8 GB | Too big | Get file → | |
| Q3_K_M | 13.3 GB | 16.0 GB | Too big | Get file → | |
| Q3_K_L | 14.3 GB | 17.0 GB | Too big | Get file → | |
| IQ4_XS | 15.2 GB | 17.9 GB | Too big | Get file → | |
| Q4_K_S | 15.6 GB | 18.3 GB | Too big | Get file → | |
| Q4_K_M | 16.5 GB | 19.2 GB | Too big | Get file → | |
| Q5_K_S | 18.7 GB | 21.4 GB | Too big | Get file → | |
| Q5_K_M | 19.2 GB | 21.9 GB | Too big | Get file → | |
| Q6_K | 22.1 GB | 24.8 GB | Too big | Get file → | |
| Q8_0 | 28.6 GB | 31.3 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_S (file size 15.59 GB, estimated 18.3 GB memory). For 12 GB VRAM, Q4_K_M (16.55 GB, 19.2 GB memory) is a good balance. For 16 GB VRAM, Q5_K_S (18.68 GB, 21.4 GB memory) offers higher quality.
What GPU do you need for GameScaling 27B S3?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: the best quant that fits is Q3_K_M (13.3 GB file, about 16.0 GB in use).
- 24 GB VRAM: the best quant that fits is Q5_K_M (19.2 GB file, about 21.9 GB in use).
- 32 GB VRAM: the best quant that fits is Q8_0 (28.6 GB file, about 31.3 GB in use).
Which GameScaling 27B S3 quant should you choose?
Q4_K_M is the usual starting point for GameScaling 27B S3: it balances file size and quality well. Here it is a 16.5 GB file.
The Q8_0 file is 28.6 GB, about 1.7 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is Q2_K at 10.7 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about GameScaling 27B S3
How much VRAM does GameScaling 27B S3 need?
About 19.2 GB at Q4_K_M with an 8K context window. That is the 16.5 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run GameScaling 27B S3 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 13.4 GB, so you would need a larger GPU or CPU offloading.
What is the file size of GameScaling 27B S3 GGUF?
It depends on the quant: Q2_K is 10.7 GB, Q3_K_S is 12.1 GB, Q3_K_M is 13.3 GB, Q3_K_L is 14.3 GB, IQ4_XS is 15.2 GB, Q4_K_S is 15.6 GB and 5 more.
What context length does GameScaling 27B S3 support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 2.2 GB at 8K and 8.7 GB at 32K.
Where can I download GameScaling 27B S3 GGUF?
The files are on Hugging Face in the mradermacher/GameScaling-27B-S3-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does GameScaling 27B S3 use?
Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run GameScaling 27B S3 on 8 GB VRAM?
Use the Q4_K_S quant; it requires about 18.3 GB memory at 8k context, which is above 8 GB, so it will not fit. A lower‑precision quant like Q2_K may fit but with higher memory usage.
Does the model support vision?
Yes, the base model is a vision‑language model and can process game frames, but some quant files may not include vision weights.
What license does the model use?
The model is released under the apache‑2.0 license.