steampunque/Qwen3-30B-A3B-Instruct-2507-MP-GGUF

Mixed Precision GGUF layer quantization of Qwen3-30B-A3B-Instruct-2507 by Qwen

Original model: https://huggingface.co/Qwen/Qwen3-30B-A3B-Instruct-2507

The hybrid quant employs different quantization levels on a per layer basis to increase flexibility of trading off performance vs file size. Less parameter bits are used at deep layers and more bits at cortex layers to simultaneously optimize quantized size and model performance. For this file the layer quants are as follows:

   LAYER_TYPES='[
   [0 ,"Q4_K_M"],[1 ,"Q4_K_M"],[2 ,"Q4_K_S"],[3 ,"Q3_K_L"],[4 ,"Q3_K_M"],[5 ,"Q3_K_M"],[6 ,"Q3_K_M"],[7 ,"Q3_K_M"],
   [8 ,"Q3_K_L"],[9 ,"Q3_K_M"],[10,"Q3_K_L"],[11,"Q3_K_M"],[12,"Q3_K_L"],[13,"Q3_K_M"],[14,"Q3_K_L"],[15,"Q3_K_M"],
   [16,"Q3_K_L"],[17,"Q3_K_M"],[18,"Q3_K_L"],[19,"Q3_K_M"],[20,"Q3_K_L"],[21,"Q3_K_L"],[22,"Q3_K_L"],[23,"Q3_K_L"],
   [24,"Q3_K_L"],[25,"Q3_K_L"],[26,"Q3_K_L"],[27,"Q3_K_L"],[28,"Q4_K_S"],[29,"Q3_K_L"],[30,"Q4_K_S"],[31,"Q3_K_L"],
   [32,"Q4_K_S"],[33,"Q3_K_L"],[34,"Q4_K_S"],[35,"Q3_K_L"],[36,"Q4_K_S"],[37,"Q4_K_S"],[38,"Q4_K_S"],[39,"Q4_K_S"],
   [40,"Q4_K_S"],[41,"Q4_K_S"],[42,"Q4_K_S"],[43,"Q4_K_S"],[44,"Q4_K_M"],[45,"Q5_K_S"],[46,"Q5_K_M"],[47,"Q6_K"  ]
   ]'
   FLAGS="--token-embedding-type Q6_K --output-tensor-type Q6_K --layer-types-high"

These layer quants were optimized for good performance on both code and reasoning problems across a small set of curated test/eval prompts and also for generation stability with greedy sampling. NOTE: this quant was re-uploaded with a different layer quant distribution after the initial upload. To verify correct file make sure its ~16.8G in size or check sha256 on the model.

Comparison:

Quant	size	PPL	Comment
IQ4_XS	16.6e9	7.4	default embed and output, unstable with greedy sampling
Q4_K_H	16.8e9	7.4	Q6_K embed Q6_K output, stable with greedy sampling

Note the straightforward IQ4_XS quant was found unusable. The model will go into infinite repetition loop at random points on some prompts with greedy sampling. This issue was not found across the eval set used to optimize the hybrid layer quants (by design).

Usage:

Compared to the first Qwen3-30B-A3B this model changes:

1. Bigger native context of 256k extendable to 1M with rope
2. No thinking mode is available, however the model can automatically generate wait ... reflections during generations depending on the problem.

This moe model can be efficiently run by offloading expert tensors to CPU via -ot exps=CPU to open up very large context space. The smaller size of the optimally quantized parameters will give an effective boost in CPU processing speed due to reducing the memory BW needed to repeatedly copy them from main memory to SIMD regs. It can also run fully offloaded on GPU via RPC or high VRAM GPU.

The recommended speculator for the model is Qwen3-0.6B if the inference platform can support vocabulary translation between draft and target. Approximate performance using 4070 GPU and a 9900k CPU with a downstream speculator used with llama.cpp:

Config	block 8 speculated code gen speed	block 4 non code gen speed
2 4070, RPC, fully offloaded to GPU	83 t/s	41 t/s
1 4070, -ot exps=CPU, CPU=9900k	34 t/s	18 t/s

Benchmarks:

Evals for the model will eventually be given here: https://huggingface.co/spaces/steampunque/benchlm.

Download the file from below:

Link	Type	Size/e9 B	Notes
Qwen3-30B-A3B-Instruct-2507.Q4_K_H.gguf	Q4_K_H	16.8e9 B	~IQ4_XS size

A discussion thread about the hybrid layer quant approach can be found here on the llama.cpp git repository:

https://github.com/ggml-org/llama.cpp/discussions/13040