Are Unsloth models as good as I read?

Has anybody done some comparing between the models that Unsloth offers and their counter part? For example: I've been using qwen3.6:35b-a3b Q4_K_M , and on my MBP 64GB I get around 39 t/s Using Unsloth Studio, unsloth/qwen3.6:35b-a3b UD-Q4_K_XL I get around 57 t/s The difference in speed is significant. From what I've understood the Unsloth model runs a per-layer sensitivity analysis and assigns different quantization levels depending on how "important" each layer is. This obviously makes the mo

More from LocalLlama

Used a Claude Code skill to fine-tune Qwen3-1.7B from 327 noisy traces, matches GLM-5

LocalLlama

We benchmarked gpt-oss-120b across 6 inference providers and found a 10x throughput spread

LocalLlama

I’m starvin’

LocalLlama

🚀Pocket LLM v1.5.0 is out: offline Android LLM chat with voice, image input, OCR, and camera capture

LocalLlama

FP4 inference in llama.cpp (NVFP4) and ik_llama.cpp (MXFP4) landed - Finally

LocalLlama

GLM 5.1 Locally: 40tps, 2000+ pp/s

LocalLlama