Kimi VL A3B Thinking 2506 needs about 11.5 GB of memory at Q4_K_M (a 10.5 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 25 quants, from 0.0 GB to 14.3 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 1.9 GB.
Kimi VL A3B Thinking 2506 – a 16B vision‑language model with multimodal reasoning
Kimi VL A3B Thinking 2506 is a 16B vision‑language model based on the deepseek2 architecture. It supports 131072 context tokens and is trained for reasoning and vision tasks. The model is released under the MIT license.
Good at
- Multimodal reasoning
- High‑resolution vision
- Video understanding
Limits
- No vision support beyond images
- Requires flash‑attn for efficient inference
- Large memory for higher‑quality quants
Best for
- Chatbots with image input
- Video analysis pipelines
- High‑resolution image captioning
Who it’s for. Developers building multimodal applications that need reasoning and vision. The model is suitable for research and production use where a 16B parameter count is acceptable.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.0 GB | 1.0 GB | Fits | Get file → | |
| IQ1_S | 5.1 GB | 6.1 GB | Fits | Get file → | |
| IQ1_M | 5.4 GB | 6.4 GB | Fits | Get file → | |
| IQ2_XXS | 5.8 GB | 6.8 GB | Fits | Get file → | |
| IQ2_XS | 6.1 GB | 7.1 GB | Fits | Get file → | |
| IQ2_S | 6.2 GB | 7.1 GB | Fits | Get file → | |
| IQ2_M | 6.5 GB | 7.5 GB | Tight | Get file → | |
| Q2_K | 6.6 GB | 7.6 GB | Tight | Get file → | |
| Q2_K_S | 6.6 GB | 7.6 GB | Tight | Get file → | |
| IQ3_XXS | 7.1 GB | 8.1 GB | Too big | Get file → | |
| IQ3_XS | 7.3 GB | 8.3 GB | Too big | Get file → | |
| IQ3_S | 7.6 GB | 8.6 GB | Too big | Get file → | |
| Q3_K_S | 7.6 GB | 8.6 GB | Too big | Get file → | |
| IQ3_M | 7.7 GB | 8.7 GB | Too big | Get file → | |
| Q3_K_M | 8.3 GB | 9.3 GB | Too big | Get file → | |
| Q3_K_L | 8.6 GB | 9.6 GB | Too big | Get file → | |
| IQ4_XS | 8.7 GB | 9.7 GB | Too big | Get file → | |
| IQ4_NL | 9.1 GB | 10.1 GB | Too big | Get file → | |
| Q4_0 | 9.1 GB | 10.1 GB | Too big | Get file → | |
| Q4_K_S | 9.7 GB | 10.7 GB | Too big | Get file → | |
| Q4_1 | 10.1 GB | 11.0 GB | Too big | Get file → | |
| Q4_K_M | 10.5 GB | 11.5 GB | Too big | Get file → | |
| Q5_K_S | 11.3 GB | 12.3 GB | Too big | Get file → | |
| Q5_K_M | 12.0 GB | 13.0 GB | Too big | Get file → | |
| Q6_K | 14.3 GB | 15.3 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (file size 5.13 GB, memory 6.1 GB). For 12 GB VRAM, IQ3_S (file size 7.65 GB, memory 8.6 GB) or Q3_K_S. For 16 GB VRAM, Q4_K_M (file size 10.54 GB, memory 11.5 GB) offers a good balance of size and quality.
What GPU do you need for Kimi VL A3B Thinking 2506?
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 the main file (0.0 GB file, about 1.0 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 1.0 GB in use).
- 8 GB VRAM: the best quant that fits is Q2_K_S (6.6 GB file, about 7.6 GB in use).
- 12 GB VRAM: the best quant that fits is Q4_K_M (10.5 GB file, about 11.5 GB in use).
- 16 GB VRAM: the best quant that fits is Q6_K (14.3 GB file, about 15.3 GB in use).
Which Kimi VL A3B Thinking 2506 quant should you choose?
Q4_K_M is the usual starting point for Kimi VL A3B Thinking 2506: it balances file size and quality well. Here it is a 10.5 GB file.
If memory is tight, the smallest file is the first one in the table at 0.0 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Kimi VL A3B Thinking 2506
How much VRAM does Kimi VL A3B Thinking 2506 need?
About 11.5 GB at Q4_K_M with an 8K context window. That is the 10.5 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 Kimi VL A3B Thinking 2506 on an 8 GB GPU?
Yes, with Q2_K_S: the 6.6 GB file needs about 7.6 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Kimi VL A3B Thinking 2506 GGUF?
It depends on the quant: file is 0.0 GB, IQ1_S is 5.1 GB, IQ1_M is 5.4 GB, IQ2_XXS is 5.8 GB, IQ2_XS is 6.1 GB, IQ2_S is 6.2 GB and 19 more.
What context length does Kimi VL A3B Thinking 2506 support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.5 GB at 8K and 1.9 GB at 32K.
Where can I download Kimi VL A3B Thinking 2506 GGUF?
The files are on Hugging Face in the mradermacher/Kimi-VL-A3B-Thinking-2506-i1-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 Kimi VL A3B Thinking 2506 use?
Hugging Face lists the license as mit. 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 Kimi VL A3B Thinking 2506 on 8 GB VRAM?
Yes, the IQ1_S quant uses 6.1 GB at 8K context and is suitable for 8 GB VRAM.
Does the model support video input?
Yes, it extends to video scenarios and performs well on VideoMMMU and Video‑MME benchmarks.
What is the recommended inference library?
VLLM is recommended for long‑decode generation up to 32K tokens, with flash‑attn installed.