community pool / model answer
LFM2.5-350M
chat · LiquidAI/LFM2.5-350M
549.9 tok/s · model median · 12 runs in the model record.
| Reference rig | Quantization | Median result | Basis | TTFT (ms) | Peak VRAM | Max context | Engines |
|---|---|---|---|---|---|---|---|
| Radeon AI Pro R9700 32GB | 4-bit | 539.6 tok/s | measured | 87 | - GB | 4,096 | hipfire |
| RTX 3060 12GB | 8-bit | 574.7 tok/s | reported | 140 | 1.9 GB | 4,608 | llama.cpp |
| RX 5700 XT 8GB | 4-bit | 257.3 tok/s | reported | 76 | - GB | 4,096 | hipfire |
| RX 6900 XT 16GB | 4-bit | 275.8 tok/s | reported | 67 | - GB | 4,096 | hipfire |
| RX 7900 XTX 24GB | 4-bit | 398.4 tok/s | reported | 57 | - GB | 4,096 | hipfire |
| RX 9070 XT 16GB | 4-bit | 479.2 tok/s | reported | 78 | - GB | 4,096 | hipfire |
| Ryzen AI Max 395 103GB | 4-bit | 343.1 tok/s | reported | 58 | - GB | 4,096 | hipfire |
Context tested
The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.