code-gen 14B params
Qwen/Qwen2.5-Coder-14B-Instruct-AWQ
Qwen/Qwen2.5-Coder-14B-Instruct-AWQ
840,207 downloads
17 likes
Hardware Recommendations
Three tiers based on VRAM requirements: minimum for basic inference, comfortable for longer contexts, headroom for future model sizes.
Minimum $599
NVIDIA GeForce RTX 4070 (12GB)
12GB VRAM 200W TDP
Runs 14B at Q4_K_M quantization with 4K context.
Comfortable $599
Apple Mac mini (M4, 16GB)
16GB unified 25W TDP
Runs 14B at Q5_K_M quantization with 8K+ context.
VRAM Requirements
| Quantization | VRAM | Use case |
|---|---|---|
| Q4_K_M | 11 GB | Basic inference, 4K context |
| Q5_K_M | 13 GB | Good quality, 8K+ context |
| Q8 | 19 GB | Near-lossless, 16K+ context |
| FP16 | 36 GB | Full precision, max quality |
Formula: params × bytes_per_param × 1.2 overhead + 2 GB base. Actual requirements vary by framework and context length.