https://gauntletbench.com/https://gauntletbench.com/about/https://gauntletbench.com/arena/https://gauntletbench.com/arena/h1/https://gauntletbench.com/arena/h2/https://gauntletbench.com/arena/h3/https://gauntletbench.com/arena/h4/https://gauntletbench.com/arena/h5/https://gauntletbench.com/arena/j1/https://gauntletbench.com/arena/j3/https://gauntletbench.com/arena/j4/https://gauntletbench.com/arena/j5/https://gauntletbench.com/arena/j6/https://gauntletbench.com/articles/https://gauntletbench.com/articles/qwen38-buyers-guide-part-1-what-compression-costs/https://gauntletbench.com/contribute/https://gauntletbench.com/education/https://gauntletbench.com/education/dense-vs-mixture-of-experts/https://gauntletbench.com/education/file-formats-gguf-mlx-safetensors/https://gauntletbench.com/education/inside-a-test-suite/https://gauntletbench.com/education/model-size-and-quantization/https://gauntletbench.com/education/reading-a-score/https://gauntletbench.com/education/reasoning-models-and-thinking/https://gauntletbench.com/education/speculative-decoding-and-mtp/https://gauntletbench.com/education/the-eight-axes-explained/https://gauntletbench.com/education/what-is-a-local-llm/https://gauntletbench.com/education/why-scores-differ-day-to-day/https://gauntletbench.com/gauntlet/https://gauntletbench.com/lab/https://gauntletbench.com/lab/m5-neural-accelerator/https://gauntletbench.com/lab/pocketpal-review/https://gauntletbench.com/lab/runtime-tracking/https://gauntletbench.com/models/bonsai-27b-ternary-2bit-mlx/https://gauntletbench.com/models/deepseek-r1-0528-qwen3-8b-4bit-mlx/https://gauntletbench.com/models/deepseek-r1-distill-llama-70b-8bit-mlx/https://gauntletbench.com/models/deepseek-r1-distill-qwen-32b-8bit-mlx/https://gauntletbench.com/models/devstral-small-2-24b-4bit-mlx/https://gauntletbench.com/models/devstral-small-2-24b-6bit-mlx/https://gauntletbench.com/models/gemma-3-1b-qat-4bit-mlx/https://gauntletbench.com/models/gemma-3-4b-qat-4bit-mlx/https://gauntletbench.com/models/gemma-3-12b-it-qat-4bit-mlx/https://gauntletbench.com/models/gemma-4-26b-a4b-8bit-mlx/https://gauntletbench.com/models/gemma-4-26b-a4b-mlx-ollama-mbp/https://gauntletbench.com/models/gemma-4-31b-it-4bit-mlx/https://gauntletbench.com/models/gemma-4-e4b-4b-8bit-mlx/https://gauntletbench.com/models/gemma-4-e4b-7.5b-q4-k-m-gguf/https://gauntletbench.com/models/gpt-oss-20b-openai-mxfp4/https://gauntletbench.com/models/gpt-oss-120b/https://gauntletbench.com/models/gpt-oss-safeguard-20b-mlx/https://gauntletbench.com/models/hermes-4-70b-4bit-mlx/https://gauntletbench.com/models/laguna-s-2.1-q4-k-m-gguf/https://gauntletbench.com/models/mistral-small-3.2-24b-8bit-mlx/https://gauntletbench.com/models/muse-glimmer-30b-gguf-kquant-17gb/https://gauntletbench.com/models/nemotron-3-nano-30b-a3b-4bit-mlx/https://gauntletbench.com/models/ornith-1.0-35b-4bit-mlx/https://gauntletbench.com/models/ornith-1.5-35b-a3b-q4-k-m-gguf/https://gauntletbench.com/models/phi-3.5-mini-instruct-4bit-mlx/https://gauntletbench.com/models/qwen3-4b-2507-non-thinking-4bit-mlx/https://gauntletbench.com/models/qwen3-coder-next-80b-4bit-mlx/https://gauntletbench.com/models/qwen3-vl-30b-a3b-instruct-4bit-mlx/https://gauntletbench.com/models/qwen3.5-122b-a10b-4bit-mlx/https://gauntletbench.com/models/qwen3.6-27b-dense-8bit-mlx/https://gauntletbench.com/models/qwen3.6-27b-fable-fusion-711-uncensored-heretic-neo-max-mtplx-8bit-mlx/https://gauntletbench.com/models/qwen3.6-27b-fable-fusion-711-uncensored-heretic-neo-max-q8-0-gguf/https://gauntletbench.com/models/qwen3.6-35b-a3b-4bit-mlx/https://gauntletbench.com/models/qwen3.6-35b-a3b-mlx-8bit-uniform-lmstudio-community/https://gauntletbench.com/models/qwen3.6-35b-a3b-unsloth-dynamic-ud-q8-k-xl-mlx-brooooooklyn/https://gauntletbench.com/models/qwen3.8-27b-gsq-rco-iq3-s-gguf-ista-daslab-ollama-mbp-identical-file-arm-of-m35/https://gauntletbench.com/models/qwen3.8-27b-mlx-4bit-lmstudio-community-ollama-mbp-identical-weights-arm-of-m25h/https://gauntletbench.com/models/qwen3.8-27b-mlx-4bit-lmstudio-community/https://gauntletbench.com/models/qwen3.8-27b-mlx-5bit-lmstudio-community/https://gauntletbench.com/models/qwen3.8-27b-mlx-6bit-lmstudio-community/https://gauntletbench.com/models/qwen3.8-27b-mlx-8bit-lmstudio-community-ollama-mbp-quant-ladder-rung/https://gauntletbench.com/models/qwen3.8-27b-mlx-8bit-lmstudio-community/https://gauntletbench.com/models/qwen3.8-27b-mtplx-optimized-speed-4bit-mlx-youssofal/https://gauntletbench.com/models/qwen3.8-27b-nvfp4-mlx-ollama-mbp/https://gauntletbench.com/models/qwen3.8-27b-q4-k-m-gguf/https://gauntletbench.com/models/qwen3.8-27b-q6-k-gguf-lmstudio-community-ollama-mbp-identical-file-arm-of-m25b/https://gauntletbench.com/models/qwen3.8-27b-q6-k-gguf/https://gauntletbench.com/models/qwen3.8-27b-q8-0-gguf/https://gauntletbench.com/models/qwen3.8-27b-ud-q6-k-xl-gguf-ollama-identical-file-control/https://gauntletbench.com/models/qwen3.8-27b-uncensored-hauhaucs-aggressive-q5-k-p-gguf-ollama-mbp-identical-file-arm-of-m29/https://gauntletbench.com/models/qwen3.8-27b-uncensored-hauhaucs-aggressive-q6-k-p-gguf-ollama-mbp-identical-file-arm-of-m27/https://gauntletbench.com/rounds/https://gauntletbench.com/rounds/browser-verification/https://gauntletbench.com/rounds/production-replay/https://gauntletbench.com/rounds/quant-ladder/https://gauntletbench.com/rounds/sideloaded-mlx-drop/https://gauntletbench.com/rounds/xhigh-minus-39/