Qwen 3.6 27B GGUF Quality Benchmark: unsloth, lmstudio-community, Jackrong, bartowski, mradermacher, ubergarm, ggml-org compared
87 GGUF quants benchmarked against BF16, ranked by KL divergence
localbench
GGUF quality benchmarks comparing unsloth, bartowski, and other uploaders. Every quant ranked by KL divergence across 250K tokens of real-world tasks, not Wikipedia.
Live Last read · last published · next check
Latest posts
Gemma 4 and Qwen 3.6 with q8_0 and q4_0 KV cache: KL divergence results
4 models tested with q8_0 and q4_0 KV cache against full-precision baseline
Qwen 3.6 35B A3B GGUF Quality Benchmark: unsloth, bartowski, lmstudio-community, ggml-org, mudler, AesSedai compared
64 GGUF quants benchmarked against BF16, ranked by KL divergence
Qwen 3.5 9B GGUF Quality Benchmark: unsloth, bartowski, lmstudio-community, mradermacher compared
72 GGUF quants benchmarked against BF16, ranked by KL divergence
Qwen 3.5 35B A3B GGUF Quality Benchmark: unsloth, bartowski, lmstudio-community, ggml-org, mradermacher, mudler, AesSedai, ubergarm compared
90 GGUF quants benchmarked against BF16, ranked by KL divergence
Qwen 3.5 27B GGUF Quality Benchmark: unsloth, bartowski, lmstudio-community, mradermacher, ubergarm compared
75 GGUF quants benchmarked against BF16, ranked by KL divergence
Gemma 4 E4B GGUF Quality Benchmark: unsloth, bartowski, lmstudio-community, ggml-org, mradermacher compared
62 GGUF quants benchmarked against BF16, ranked by KL divergence
Gemma 4 26B A4B GGUF Quality Benchmark: unsloth, bartowski, lmstudio-community, ggml-org, mradermacher, mudler compared
80 GGUF quants benchmarked against BF16, ranked by KL divergence
GGUF Quantization Quality Benchmark: Methodology
What is KL divergence? KL divergence measures how different the quantized model’s token probability distribution is from the original model’s. A KL of 0 means the quant is identical to the original. Higher values mean more information is lost. Unlike perplexity, KL divergence directly compares two models against each other. It answers the question: “how much does this quant change the model’s…
Gemma 4 31B GGUF Quality Benchmark: unsloth, bartowski, lmstudio-community, ggml-org compared
53 GGUF quants benchmarked against BF16, ranked by KL divergence