Optimized for Apple Silicon.
AI & ML interests
building world's fastest retrieval and inference engines.
Recent Activity
Small o1 at home.
King of Visual Function Calling. Generates code to interact with images and self-corrects based on rendered output.
One of the most goated models in OSS
Execution-driven RL for unmatched perf.
Delivers local performance that challenges Claude 4 Sonnet. Best for Agentic coding.
Pure SSM with no attention. Linear scaling enables constant memory cost for massive files.
Runs 80B intelligence at 3B compute footprint with 10x faster inference.
-
srswti/Qwen3-Next-80B-A3B-Thinking-8bit
Text Generation • 80B • Updated • 15 -
srswti/Qwen3-Next-80B-A3B-Thinking-6bit
Text Generation • 80B • Updated • 16 -
srswti/Qwen3-Next-80B-A3B-Thinking-4bit
Text Generation • 80B • Updated • 19 -
srswti/Qwen3-Next-80B-A3B-Instruct-8bit
Text Generation • 80B • Updated • 14
It is a good chat model.
-
srswti/Llama-3.2-11B-Vision-Instruct-abliterated
Image-to-Text • 11B • Updated • 126 -
srswti/Llama-3.2-11B-Vision-Instruct-abliterated-8-bit
Image-to-Text • 3B • Updated • 234 -
srswti/Llama-3.2-11B-Vision-Instruct-abliterated-4-bit
Image-to-Text • 2B • Updated • 345 -
srswti/Llama-3.2-3B-Instruct
Text Generation • 3B • Updated • 8
Optimized for Apple Silicon.
Delivers local performance that challenges Claude 4 Sonnet. Best for Agentic coding.
Small o1 at home.
Pure SSM with no attention. Linear scaling enables constant memory cost for massive files.
King of Visual Function Calling. Generates code to interact with images and self-corrects based on rendered output.
Runs 80B intelligence at 3B compute footprint with 10x faster inference.
-
srswti/Qwen3-Next-80B-A3B-Thinking-8bit
Text Generation • 80B • Updated • 15 -
srswti/Qwen3-Next-80B-A3B-Thinking-6bit
Text Generation • 80B • Updated • 16 -
srswti/Qwen3-Next-80B-A3B-Thinking-4bit
Text Generation • 80B • Updated • 19 -
srswti/Qwen3-Next-80B-A3B-Instruct-8bit
Text Generation • 80B • Updated • 14
One of the most goated models in OSS
It is a good chat model.
-
srswti/Llama-3.2-11B-Vision-Instruct-abliterated
Image-to-Text • 11B • Updated • 126 -
srswti/Llama-3.2-11B-Vision-Instruct-abliterated-8-bit
Image-to-Text • 3B • Updated • 234 -
srswti/Llama-3.2-11B-Vision-Instruct-abliterated-4-bit
Image-to-Text • 2B • Updated • 345 -
srswti/Llama-3.2-3B-Instruct
Text Generation • 3B • Updated • 8
Execution-driven RL for unmatched perf.