RTX 4070 Ti SUPER × Qwen2.5 14B Instruct
Qwen2.5 14B Instruct on RTX 4070 Ti SUPER
53.6best gen tok/s
2,526best prompt tok/s
4,352max context tested
7results
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 | 53.6 | 2,526 | LocalScore ↗ | 512+128 | 521 ms | llamafile 0.9.2 |
| Q4_K - Medium | 47.3 | 2,382 | LocalScore ↗ | 512+128 | 550 ms | llamafile 0.9.2 |
| Q4_K - Medium | 40.3 | 2,367 | LocalScore ↗ | 4,352 | 548 ms | llamafile 0.9.3 |
| Q4_K - Medium | 36.1 | 2,131 | LocalScore ↗ | 512+128 | 620 ms | llamafile 0.9.3 |
| Q4_K - Medium | 27.2 | 2,045 | LocalScore ↗ | 512+128 | 631 ms | llamafile 0.9.2 |
| Q4_K - Medium | 26.7 | 1,737 | LocalScore ↗ | 512+128 | 734 ms | llamafile 0.9.2 |
| Q4_K - Medium | 26.6 | 1,840 | LocalScore ↗ | 512+128 | 716 ms | llamafile 0.9.2 |
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 4070 Ti SUPER 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.