RTX 5070 Ti × Qwen2.5 14B Instruct
Qwen2.5 14B Instruct on RTX 5070 Ti
47.7best gen tok/s
2,490best prompt tok/s
4,352max context tested
12results
Measured · all numbers below come from the linked sources
llama.cpp
| Quant | Gen tok/s | Prompt tok/s | Source | Context tested | TTFT | Build |
|---|---|---|---|---|---|---|
| Q4_K - Medium | 47.7 | 2,317 | LocalScore ↗ | 512+128 | 859 ms | llamafile 0.9.2 |
| Q4_K - Medium | 41.2 | 2,490 | LocalScore ↗ | 512+128 | 514 ms | llamafile 0.9.3 |
| Q4_K - Medium | 36.5 | 2,367 | LocalScore ↗ | 512+128 | 540 ms | llamafile 0.9.3 |
| Q4_K - Medium | 31.1 | 2,195 | LocalScore ↗ | 512+128 | 586 ms | llamafile 0.9.3 |
| Q4_K - Medium | 27.6 | 2,261 | LocalScore ↗ | 512+128 | 562 ms | llamafile 0.9.2 |
| Q4_K - Medium | 26.9 | 2,262 | LocalScore ↗ | 512+128 | 563 ms | llamafile 0.9.2 |
| Q4_K - Medium | 23.2 | 1,482 | LocalScore ↗ | 512+128 | 889 ms | llamafile 0.9.2 |
| Q4_K - Medium | 20.9 | 1,129 | LocalScore ↗ | 512+128 | 1,127 ms | llamafile 0.9.3 |
| Q4_K - Medium | 20.0 | 1,127 | LocalScore ↗ | 512+128 | 1,132 ms | llamafile 0.9.3 |
| Q4_K - Medium | 20.0 | 1,128 | LocalScore ↗ | 512+128 | 1,130 ms | llamafile 0.9.3 |
| Q4_K - Medium | 14.2 | 1,740 | LocalScore ↗ | 4,352 | 730 ms | llamafile 0.9.3 |
| Q4_K - Medium | 12.6 | 1,638 | LocalScore ↗ | 4,352 | 770 ms | llamafile 0.9.3 |
Ollama
No measured Ollama results yet. Submit yours →
MLX
No measured MLX results yet. Submit yours →
"Context tested" is the largest prompt+generation length actually benchmarked (tokens), not a memory-fit estimate.
What else can RTX 5070 Ti run? →
Includes results from LocalScore, a Mozilla Builders project by cjpais, mirrored in the cn0303/local-llm-speed-benchmarks dataset. LocalScore runs use llamafile, which is built on llama.cpp.