Model files for Windows PC, RTX 4070 12 GB
Compare hosted file sizes with Windows PC, RTX 4070 12 GB memory. File size is an estimate, not a runtime guarantee.
Updated
Each file below is no larger than 60% of the listed device memory. Text-generation speed is a bandwidth estimate, not a measured run.
- Device
- Windows PC, RTX 4070 12 GB
- Device memory
- 12 GB (VRAM)
- Profile bandwidth
- 504 GB/s
4 file-size matches in this sample.
Chat and text generation
| Model | Selected file | Text speed | Next step |
|---|---|---|---|
| 0xSojalSec/MiniCPM5-2B-Abliterated-Uncensored-GGUF minicpm5-2b-abliterated-Q4_K_M.gguf | 1.45 GiB | Estimated: 71.0 tok/s | |
| 0xzknw/Granite-4.2-3B-Heretic-NX-PRIME-GGUF Granite-4.2-3B-Heretic-NX-PRIME-Q4_K_M.gguf | 2.09 GiB | Estimated: 49.4 tok/s | |
| 0xSojalSec/Abliterated-MiMo-V2.6-Distill-Qwen-9B-GGUF-MLX MiMo-V2.6-Distill-Qwen-9B-Abliterated-Q4_K_M.gguf | 5.24 GiB | Estimated: 19.7 tok/s |
Classification and tagging
| Model | Selected file | Text speed | Next step |
|---|---|---|---|
| 1bit-MONSTER/Zamba2-1.2B-instruct-GGUF Zamba2-1.2B-instruct-Q8_0.gguf | 1.71 GiB | Speed not available |
File size does not confirm that a full model fits or runs. Speed estimates use the device’s memory bandwidth and recorded model file sizes.