Solidity-LLM
by Chain-GPT
2.8B params · text-generation · 1.6k likes · 21 downloads
Solidity-LLM is a 2.8B parameter model. At Q4 quantization it requires 1GB of VRAM. It runs comfortably on GeForce RTX 4090 (472 tok/s), GeForce RTX 5090 (708 tok/s), M4 Max 128GB (172 tok/s).
GPU compatibility
| GPU | VRAM | Q4 Decode | Verdict |
|---|---|---|---|
| GeForce RTX 4090 | 24GB | 472 tok/s | comfortable |
| GeForce RTX 5090 | 32GB | 708 tok/s | comfortable |
| M4 Max 128GB | 128GB | 172 tok/s | comfortable |
| M4 Pro 48GB | 48GB | 86 tok/s | comfortable |
| M4 Pro 24GB | 24GB | 86 tok/s | comfortable |
| A100 PCIe 80 GB | 80GB | 865 tok/s | comfortable |
| H100 SXM5 80 GB | 80GB | 1740 tok/s | comfortable |
| GeForce RTX 3090 | 24GB | 417 tok/s | comfortable |
| Radeon RX 7900 XTX | 24GB | 345 tok/s | comfortable |
| GeForce RTX 4080 | 16GB | 335 tok/s | comfortable |