LAUNCHER: máš to pomalé?
random z netu
Model Name Quantization Memory Required (GB) Max ctx length Recommended Hardware Examples
Qwen3-0.6B Q4_K_M ~0.5 32K Virtually any modern PC or Mac; integrated graphics are sufficient. Mobile phones
Qwen3-1.7B Q4_K_M ~1.3 32K Any modern system with a discrete or recent integrated GPU; basic Apple Silicon Macs (M1/M2/M3/M4).
Qwen3-4B Q4_K_M ~2.5 32K GPUs with >= 4GB VRAM (e.g., older GTX cards, RX series); entry-level Apple Silicon.
Qwen3-8B Q4_K_M ~5.0 128K GPUs with >= 8GB VRAM (e.g., RTX 3050, RX 6600); standard Apple Silicon chips (M1/M2/M3/M4 base models).
Qwen3-14B Q4_K_M ~9.0 128K GPUs with >= 12GB VRAM (RTX 3060 12GB, RTX 4060 Ti 16GB, potential RTX 5060 Ti 16GB); Apple Silicon Pro/Max chips (M1/M2/M3/M4 Pro/Max with >= 16GB unified memory).
Qwen3-30B Q4_K_M ~18.6 128K Single GPUs: Used RTX 3090 (24GB), P40 (24GB, low bandwidth), L4 (24GB, low bandwidth), A10 (24GB), Tesla V100 32GB. Multi-GPU: 2x RTX 3060 12GB, 2x 4060 Ti 16GB. Apple Silicon: M-series Pro/Max/Ultra with >= 24GB unified memory. PC with fast DDR5 RAM (see performance notes).
Qwen3-32B Q4_K_M ~19.8 128K Single GPUs: Used RTX 3090 (24GB), P40, L4, A10, V100 32GB. Multi-GPU: 2x RTX 3060 12GB, 2x 4060 Ti 16GB. Apple Silicon: M-series Pro/Max/Ultra with >= 24GB unified memory. PC with fast DDR5 RAM (32GB+ recommended).
Qwen3-235B* Q3_K_L ~112 128K Multi-GPU: 5x RTX 3090 (120GB), 3x L40/A40 (144GB), 4x V100 32GB (128GB), 2x RTX Pro 6000 Ada (96GB – too small). Apple Silicon: Mac Studio/Pro M-Ultra with >= 128GB unified memory. PC Workstation/Server: 128GB+ high-speed RAM (DDR5 preferred).