YOUR AI. YOUR DEVICE.
Models & engines
Choose how Gondel Nova thinks. No metered AI calls, no automatic cloud fallback.
This device
No compatible GPU adapter is available on this device.
WebGPU: Not supportedSystem memory: 4 GB (reported)Free browser quota: Unknown
Compatible Qwen catalogue
WebLLM 0.2.85Only builds included in the installed runtime are listed. Memory estimates are runtime VRAM requirements, not download sizes; loading is the final test.
Qwen3-0.6B
Qwen family · 4-bit quantized · shader-f16
~1403.34 MB GPU memoryNot downloaded
Qwen2.5-0.5B-Instruct
Qwen family · 4-bit quantized · shader-f16
~944.62 MB GPU memoryNot downloaded
Qwen2.5-1.5B-Instruct
Qwen family · 4-bit quantized · shader-f16
~1629.75 MB GPU memoryNot downloaded
Downloads happen only with your approval. Assets come from MLC's model hosts; prompts stay local. GPU, browser memory, storage eviction and model licensing still apply. No browser or phone is guaranteed compatible.