divinetribe/Nemotron-3-Nano-Omni-30B-Abliterated-MM-bf16
Image-Text-to-Text • 33B • Updated • 86
Multimodal Abliterated (MLX) — Vision + Audio
Uncensored MLX models for Apple Silicon that see (and one that hears). Language stack abliterated, vision/audio towers intact.
Note tri-modal: text + vision + audio, full bf16. ~62 gb. the flagship.
divinetribe/Nemotron-3-Nano-Omni-30B-Abliterated-MM-8bit
Image-Text-to-Text • 10B • Updated • 55Note tri-modal, 8-bit. ~33 gb. the daily driver.
divinetribe/Nemotron-3-Nano-Omni-30B-Abliterated-MM-4bit
Image-Text-to-Text • 6B • Updated • 68Note tri-modal, 4-bit. ~18 gb. fits a 32 gb mac.
divinetribe/Muse-Glimmer-30B-Abliterated-MM-bf16
Image-Text-to-Text • 30B • Updated • 303Note vision, full bf16. reads text out of photos. ~55 gb.
divinetribe/Huihui-Qwen3-VL-32B-Instruct-abliterated-4bit-mlx
Image-Text-to-Text • 6B • Updated • 332 • 1Note vision, 4-bit. deepstack tower, good at fine detail. ~19 gb.
divinetribe/gemma-4-12B-it-abliterated-4bit-mlx
Text Generation • 3B • Updated • 107Note vision, 4-bit. the smallest multimodal here. ~11 gb.