32gb : tune

presets : add 32GB

presets : split

preset : add pi to the harnesses

presets : fix reasoning end message

presets : swithc to default sampling + 4-bit pure drafters

presets : add MTP

presets : add draft spec decoding

presets : preserve thinking

presets : add qwen3.6 27B

presets : add gemma 4 + qwen 3.6

presets : add qwen3.5

presets : add Q3CN

presets : bump top-p for GLM 4.7

presets : add GLM4.7 flash without thinking

preset : add glm-4.7-flash-hf

presets : add qwen3 next

preset : more models

preset : add models

preset : add reasoning modes

preset : fix ubatch param

preset : fix typo

preset : fix batch and ubatch names

preset : extend context + comments

preset : update aliases to have "-hf" suffixes

Update preset.ini

Upload preset.ini

initial commit