cognition ai Kevin 32B needs about 22.5 GB of memory at Q4_K_M (a 19.9 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 27 quants, from 9.0 GB (IQ2_XXS) to 65.5 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.
Kevin 32B – a 32.8B‑parameter model for efficient CUDA kernel generation
Kevin 32B is a 32.8B‑parameter model fine‑tuned to write efficient CUDA kernels. It was trained with multi‑turn reinforcement learning and benchmarked on KernelBench. The model is available in multiple GGUF quantizations for llama.cpp usage.
Good at
- CUDA kernel generation
- Efficient code synthesis
- KernelBench performance
Limits
- No vision support
- Large context size requires high VRAM
- License not specified
Best for
- Developers writing CUDA kernels
- Researchers benchmarking GPU code
- Educational tools for GPU programming
Who it’s for. Engineers and researchers who need a specialized LLM for CUDA kernel development. The model is suited for those who can allocate sufficient VRAM and use llama.cpp or compatible projects.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| IQ2_XXS | 9.0 GB | 11.7 GB | Too big | Get file → | |
| IQ2_XS | 10.0 GB | 12.6 GB | Too big | Get file → | |
| IQ2_S | 10.4 GB | 13.0 GB | Too big | Get file → | |
| IQ2_M | 11.3 GB | 13.9 GB | Too big | Get file → | |
| Q2_K | 12.3 GB | 15.0 GB | Too big | Get file → | |
| IQ3_XXS | 12.8 GB | 15.5 GB | Too big | Get file → | |
| Q2_K_L | 13.1 GB | 15.7 GB | Too big | Get file → | |
| IQ3_XS | 13.7 GB | 16.4 GB | Too big | Get file → | |
| Q3_K_S | 14.4 GB | 17.0 GB | Too big | Get file → | |
| IQ3_M | 14.8 GB | 17.5 GB | Too big | Get file → | |
| Q3_K_M | 15.9 GB | 18.6 GB | Too big | Get file → | |
| Q3_K_L | 17.2 GB | 19.9 GB | Too big | Get file → | |
| IQ4_XS | 17.7 GB | 20.3 GB | Too big | Get file → | |
| Q3_K_XL | 17.9 GB | 20.6 GB | Too big | Get file → | |
| IQ4_NL | 18.7 GB | 21.3 GB | Too big | Get file → | |
| Q4_0 | 18.7 GB | 21.4 GB | Too big | Get file → | |
| Q4_K_S | 18.8 GB | 21.4 GB | Too big | Get file → | |
| Q4_K_M | 19.9 GB | 22.5 GB | Too big | Get file → | |
| Q4_K_L | 20.4 GB | 23.1 GB | Too big | Get file → | |
| Q4_1 | 20.6 GB | 23.3 GB | Too big | Get file → | |
| Q5_K_S | 22.6 GB | 25.3 GB | Too big | Get file → | |
| Q5_K_M | 23.3 GB | 25.9 GB | Too big | Get file → | |
| Q5_K_L | 23.7 GB | 26.4 GB | Too big | Get file → | |
| Q6_K | 26.9 GB | 29.5 GB | Too big | Get file → | |
| Q6_K_L | 27.3 GB | 29.9 GB | Too big | Get file → | |
| Q8_0 | 34.8 GB | 37.5 GB | Too big | Get file → | |
| BF16 | 65.5 GB (2 parts) | 68.2 GB | Too big | Get file → |
For 8 GB VRAM, use IQ2_XXS (9.03 GB file, ~11.7 GB memory). For 16 GB VRAM, IQ4_XS (17.69 GB file, ~20.3 GB memory) or Q4_K_M (19.85 GB file, ~22.5 GB memory). For 32 GB VRAM, Q5_K_L (23.74 GB file, ~26.4 GB memory) or Q6_K (26.89 GB file, ~29.5 GB memory). Estimates are approximate.
What GPU do you need for cognition ai Kevin 32B?
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: the best quant that fits is IQ2_XXS (9.0 GB file, about 11.7 GB in use).
- 16 GB VRAM: the best quant that fits is Q2_K_L (13.1 GB file, about 15.7 GB in use).
- 24 GB VRAM: the best quant that fits is Q4_1 (20.6 GB file, about 23.3 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K_L (27.3 GB file, about 29.9 GB in use).
- 48 GB VRAM: the best quant that fits is Q8_0 (34.8 GB file, about 37.5 GB in use).
Which cognition ai Kevin 32B quant should you choose?
Q4_K_M is the usual starting point for cognition ai Kevin 32B: it balances file size and quality well. Here it is a 19.9 GB file.
The Q8_0 file is 34.8 GB, about 1.8 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_XXS at 9.0 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 cognition ai Kevin 32B
How much VRAM does cognition ai Kevin 32B need?
About 22.5 GB at Q4_K_M with an 8K context window. That is the 19.9 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 cognition ai Kevin 32B on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 11.7 GB, so you would need a larger GPU or CPU offloading.
What is the file size of cognition ai Kevin 32B GGUF?
It depends on the quant: IQ2_XXS is 9.0 GB, IQ2_XS is 10.0 GB, IQ2_S is 10.4 GB, IQ2_M is 11.3 GB, Q2_K is 12.3 GB, IQ3_XXS is 12.8 GB and 21 more.
What context length does cognition ai Kevin 32B support?
The GGUF header lists a trained context length of 40,960 tokens (40K). The KV cache grows with the context you use: about 2.1 GB at 8K and 8.6 GB at 32K.
Where can I download cognition ai Kevin 32B GGUF?
The files are on Hugging Face in the bartowski/cognition-ai_Kevin-32B-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 Kevin 32B on 8 GB VRAM?
Use the IQ2_XXS quantization. It requires about 11.7 GB of VRAM at 8 k context, which is close to the limit of 8 GB but may work with lower context.
Does Kevin 32B support vision tasks?
No, the model has no vision capabilities.
What is the context length of Kevin 32B?
The trained context length is 40960 tokens.