Qolda needs about 4.2 GB of memory at Q4_K_M (a 2.5 GB file) with an 8K context window, so it fits on a 6 GB GPU. This repository has 19 quants, from 0.3 GB (IQ2_M) to 8.1 GB (F16). Longer context needs more: at 32K tokens the KV cache alone adds about 4.8 GB.
Qolda – 4.02B parameter vision‑language model for Kazakh, Russian, and English
Qolda is a 4.02B parameter model built on qwen3 architecture. It combines InternViT‑300M vision encoder with Qwen3‑4B language model. The model supports reasoning, tools, and vision tasks.
Good at
- Vision‑language reasoning
- Tool usage
- Multilingual support
Limits
- No vision support on CPU‑only setups
- Requires at least 8k context for accurate memory estimates
- License not specified
Best for
- Generating captions for images
- Answering multilingual queries
- Assisting in visual data analysis
Who it’s for. Developers building multilingual vision‑language applications. Researchers needing a compact 4B model with vision capabilities. Teams requiring a model that can run on GPUs with 8k context support.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ2_M | 0.3 GB | 2.0 GB | Fits | Get file → | |
| IQ3_XXS | 0.3 GB | 2.0 GB | Fits | Get file → | |
| TQ1_0 | 1.1 GB | 2.8 GB | Fits | Get file → | |
| Q2_K | 1.7 GB | 3.4 GB | Fits | Get file → | |
| Q3_K_S | 1.9 GB | 3.6 GB | Fits | Get file → | |
| Q3_K_M | 2.1 GB | 3.8 GB | Fits | Get file → | |
| IQ4_XS | 2.3 GB | 4.0 GB | Fits | Get file → | |
| Q4_0 | 2.4 GB | 4.1 GB | Fits | Get file → | |
| Q4_K_S | 2.4 GB | 4.1 GB | Fits | Get file → | |
| IQ4_NL | 2.4 GB | 4.1 GB | Fits | Get file → | |
| Q4_K_M | 2.5 GB | 4.2 GB | Fits | Get file → | |
| Q4_1 | 2.6 GB | 4.3 GB | Fits | Get file → | |
| Q5_K_S | 2.8 GB | 4.5 GB | Fits | Get file → | |
| Q5_K_M | 2.9 GB | 4.6 GB | Fits | Get file → | |
| Q6_K | 3.3 GB | 5.0 GB | Fits | Get file → | |
| Q8_0 | 4.3 GB | 6.0 GB | Fits | Get file → | |
| Q8_0 | 4.3 GB | 6.0 GB | Fits | Get file → | |
| BF16 | 8.1 GB | 9.8 GB | Too big | Get file → | |
| F16 | 8.1 GB | 9.8 GB | Too big | Get file → |
For 8GB VRAM, use IQ4_XS (file size 2.29GB, memory 4GB) or Q4_K_M (2.5GB, 4.2GB). For 16GB VRAM, Q5_K_M (2.89GB, 4.6GB) or Q6_K (3.31GB, 5GB) are suitable. Larger VRAM allows Q8_0 (4.28GB, 6GB) for full precision.
What GPU do you need for Qolda?
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: the best quant that fits is IQ4_XS (2.3 GB file, about 4.0 GB in use).
- 6 GB VRAM: the best quant that fits is Q8_0 (4.3 GB file, about 6.0 GB in use).
- 8 GB VRAM: the best quant that fits is Q8_0 (4.3 GB file, about 6.0 GB in use).
- 12 GB VRAM: the best quant that fits is BF16 (8.1 GB file, about 9.8 GB in use).
Which Qolda quant should you choose?
Q4_K_M is the usual starting point for Qolda: it balances file size and quality well. Here it is a 2.5 GB file.
The Q8_0 file is 4.3 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 IQ2_M at 0.3 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
Frequently asked questions about Qolda
How much VRAM does Qolda need?
About 4.2 GB at Q4_K_M with an 8K context window. That is the 2.5 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 Qolda on an 8 GB GPU?
Yes, with Q8_0: the 4.3 GB file needs about 6.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Qolda GGUF?
It depends on the quant: IQ2_M is 0.3 GB, IQ3_XXS is 0.3 GB, TQ1_0 is 1.1 GB, Q2_K is 1.7 GB, Q3_K_S is 1.9 GB, Q3_K_M is 2.1 GB and 13 more.
What context length does Qolda support?
The GGUF header lists a trained context length of 40,960 tokens (40K). The KV cache grows with the context you use: about 1.2 GB at 8K and 4.8 GB at 32K.
Where can I download Qolda GGUF?
The files are on Hugging Face in the issai/Qolda_GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
Can I run Qolda on 8GB VRAM?
Yes, use IQ4_XS or Q4_K_M quantizations, which estimate 4GB memory at 8k context.
Does Qolda support vision tasks on CPU?
No, vision support requires GPU backend; CPU-only inference is not supported.
What context length does Qolda need?
It was trained with 40960 tokens; 8k context is the minimum for accurate memory estimates.