PRISM Qwen3.6 for Mac (MLX/oMLX) and NVIDIA Blackwell NVFP4. One-click Mac setup + GPU vLLM.
Ex0bit PRO
Ex0bit
AI & ML interests
None yet
Recent Activity
new activity 17 days ago
Ex0bit/Qwen3.6-27B-PRISM-PRO-DQ:MLX version for Qwen3.6 Models? updated a collection 17 days ago
Qwen3.6 PRISM updated a collection 17 days ago
Qwen3.6 PRISMOrganizations
PRISM abliterated variants of Qwen 3.5 122B-A10B — multimodal MoE with hybrid attention. PRO, LITE, and GGUF formats.
PRISM abliterated MiniMax M2.1 and M2.5 MoE — reasoning, coding, agentic. GGUF and safetensors.
PRISM abliterated StepFun Step 3.5 Flash MoE — safetensors, GGUF, NVFP4.
Allen AI OLMo 3 — abliterated 7B/32B-Think and NVFP4 quantized with 1M context.
Research papers and experimental models.
NVFP4-quantized Qwen 3 VLTO models for NVIDIA Blackwell / DGX Spark.
PRISM abliterated GLM models — GLM-4.6V vision, GLM-4.7, GLM-4.7-Flash MoE.
PRISM abliterated Moonshot Kimi K2.5 — multimodal MoE with REAP expert pruning.
PRISM abliterated NVIDIA Nemotron 3 Nano 30B-A3B — Mamba-2 hybrid MoE.
Experimental and research model variants.
PRISM Qwen3.6 for Mac (MLX/oMLX) and NVIDIA Blackwell NVFP4. One-click Mac setup + GPU vLLM.
PRISM abliterated variants of Qwen 3.5 122B-A10B — multimodal MoE with hybrid attention. PRO, LITE, and GGUF formats.
NVFP4-quantized Qwen 3 VLTO models for NVIDIA Blackwell / DGX Spark.
PRISM abliterated MiniMax M2.1 and M2.5 MoE — reasoning, coding, agentic. GGUF and safetensors.
PRISM abliterated GLM models — GLM-4.6V vision, GLM-4.7, GLM-4.7-Flash MoE.
PRISM abliterated StepFun Step 3.5 Flash MoE — safetensors, GGUF, NVFP4.
PRISM abliterated Moonshot Kimi K2.5 — multimodal MoE with REAP expert pruning.
Allen AI OLMo 3 — abliterated 7B/32B-Think and NVFP4 quantized with 1M context.
PRISM abliterated NVIDIA Nemotron 3 Nano 30B-A3B — Mamba-2 hybrid MoE.
Research papers and experimental models.
Experimental and research model variants.