KE

KernelGen LM 32B

KernelgenGGUF by mradermacherUpdated 2026-05-19
Download on Hugging Face
Parameters32.8B
Context40K
Layers64
Architectureqwen3
Smallest quant0.0 GB
Quick answer

KernelGen LM 32B needs about 22.4 GB of memory at Q4_K_M (a 19.8 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 24 quants, from 0.0 GB to 26.9 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.

KernelGen LM 32B – 32.8B parameter model for Ascend NPU kernel generation

KernelGen LM 32B is a 32.8B parameter LLM built on Qwen3 architecture, fine‑tuned for low‑level NPU kernel generation on Huawei Ascend hardware. It supports coding, reasoning, and tool use, and is released under Apache‑2.0.

Good at

  • Low‑level NPU kernel generation
  • Coding with AscendC
  • Reasoning for hardware‑specific tasks

Limits

  • No vision support
  • Requires large VRAM for full precision
  • License is Apache‑2.0

Best for

  • Generating AscendC kernels
  • Fine‑tuning for hardware‑specific code
  • Evaluating kernel correctness on Ascend hardware

Who it’s for. Developers and researchers working on Huawei Ascend NPUs who need a domain‑specific LLM for kernel generation and evaluation. The model is suitable for teams that can run large‑scale inference and have access to Ascend hardware for testing.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
—0.0 GB 2.7 GB
Fits Get file →
IQ1_S7.3 GB 10.0 GB
Too big Get file →
IQ1_M8.0 GB 10.6 GB
Too big Get file →
IQ2_XXS9.0 GB 11.7 GB
Too big Get file →
IQ2_XS10.0 GB 12.6 GB
Too big Get file →
IQ2_S10.5 GB 13.2 GB
Too big Get file →
IQ2_M11.4 GB 14.0 GB
Too big Get file →
Q2_K_S11.5 GB 14.1 GB
Too big Get file →
Q2_K12.3 GB 15.0 GB
Too big Get file →
IQ3_XXS12.8 GB 15.5 GB
Too big Get file →
IQ3_XS13.7 GB 16.4 GB
Too big Get file →
Q3_K_S14.4 GB 17.0 GB
Too big Get file →
IQ3_S14.4 GB 17.1 GB
Too big Get file →
IQ3_M14.9 GB 17.6 GB
Too big Get file →
Q3_K_M16.0 GB 18.6 GB
Too big Get file →
Q3_K_L17.3 GB 20.0 GB
Too big Get file →
IQ4_XS17.7 GB 20.3 GB
Too big Get file →
Q4_018.7 GB 21.4 GB
Too big Get file →
Q4_K_S18.8 GB 21.4 GB
Too big Get file →
Q4_K_M19.8 GB 22.4 GB
Too big Get file →
Q4_120.6 GB 23.3 GB
Too big Get file →
Q5_K_S22.6 GB 25.3 GB
Too big Get file →
Q5_K_M23.2 GB 25.9 GB
Too big Get file →
Q6_K26.9 GB 29.5 GB
Too big Get file →

For 8 GB VRAM, use IQ1_S (file size 7.32 GB, memory 10 GB) or IQ1_M (7.96 GB, 10.6 GB). For 12 GB VRAM, IQ2_XS (9.95 GB, 12.6 GB) or Q2_K (12.34 GB, 15 GB) are suitable. Larger VRAM allows higher‑quality quants such as Q4_K_M (19.76 GB, 22.4 GB).

What GPU do you need for KernelGen LM 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: the best quant that fits is the main file (0.0 GB file, about 2.7 GB in use).
  • 6 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 2.7 GB in use).
  • 8 GB VRAM: the best quant that fits is the main file (0.0 GB file, about 2.7 GB in use).
  • 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 IQ3_XXS (12.8 GB file, about 15.5 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 (26.9 GB file, about 29.5 GB in use).

Which KernelGen LM 32B quant should you choose?

Q4_K_M is the usual starting point for KernelGen LM 32B: it balances file size and quality well. Here it is a 19.8 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

Architectureqwen3
Parameters32.8B
Layers64
Trained context40,960 tokens
KV cache (fp16)256 KB per token

Source

Quantized bymradermacher
FamilyKernelgen
Licenseapache-2.0
Downloads (HF)1,015
Last updated2026-05-19

Frequently asked questions about KernelGen LM 32B

How much VRAM does KernelGen LM 32B need?

About 22.4 GB at Q4_K_M with an 8K context window. That is the 19.8 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 KernelGen LM 32B on an 8 GB GPU?

Yes, with the smaller files: the 0.0 GB file needs about 2.7 GB at 8K context. Larger quants in this repository do not fit in 8 GB.

What is the file size of KernelGen LM 32B GGUF?

It depends on the quant: file is 0.0 GB, IQ1_S is 7.3 GB, IQ1_M is 8.0 GB, IQ2_XXS is 9.0 GB, IQ2_XS is 10.0 GB, IQ2_S is 10.5 GB and 18 more.

What context length does KernelGen LM 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 KernelGen LM 32B GGUF?

The files are on Hugging Face in the mradermacher/KernelGen-LM-32B-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 KernelGen LM 32B 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 KernelGen LM 32B on 8 GB VRAM?

The smallest quant for 8 GB VRAM is IQ1_S, which uses 10 GB of memory at 8 k context. It will not fit on 8 GB VRAM without off‑loading.

Does this model support vision tasks?

No, the model has no vision capabilities.

What license applies to this model?

The model is released under the Apache‑2.0 license.

Other GGUFs of KernelGen-LM-32B

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Kernelgen models

All Kernelgen →

Other medium models

All medium models →