Explore benchmarks
Filter by model, hardware, workload and quantization. Every score links to the comparisons and experiments behind it.
Top Engines
details
Speed Decode speed: tokens generated per second.
Score: 100 is the best result in every comparison an engine entered; 80 is 80% of the best, on average. Only complete experiments with a passed output check count. How rankings work
The score combines counted comparisons. Select one comparison to see its measured values.
† ExecuTorch: approximate comparison. The experiments use different effective KV cache precision: fp32, f16, uzu default. The experiments here use different 4-bit schemes: 8da4w, Q4_K_M, mirai-m. The experiments use different effective KV cache precision: fp32, litert-lm default, f16. The experiments here use different 4-bit schemes: 8da4w, litert-int4, Q4_K_M.
† llama.cpp: approximate comparison. The experiments use different effective KV cache precision: fp32, f16, uzu default. The experiments here use different 4-bit schemes: 8da4w, Q4_K_M, mirai-m. The experiments use different effective KV cache precision: fp32, litert-lm default, f16. The experiments here use different 4-bit schemes: 8da4w, litert-int4, Q4_K_M.
uzu: Results in 1 of 4 comparisons; a rank needs 2.
LiteRT-LM: Its output check failed: 9/12 of the checks the model passes at BF16; needs 11 and the mandatory ones (needle-start, needle-end, continuation-128). Left out, failed by every BF16 configuration: none.
Coverage
| Desktop | Mobile | |||
|---|---|---|---|---|
| M5 Max 64 GiB macOS 27 Apple M5 Max | OnePlus 13 16 GB Android 16 Qualcomm Snapdragon 8 Elite (SM8750) | Pixel 8 8 GB Android 16 Google Tensor G3 | iPhone 12 4 GB iOS 27 Apple A14 Bionic | |
| ExecuTorch | 872/2 out of 100, in 2 of 2 comparisons | 1002/2 out of 100, in 2 of 2 comparisons | 1001/1 out of 100, in 1 of 1 comparisons | 941/1 out of 100, in 1 of 1 comparisons |
| llama.cpp | 822/2 out of 100, in 2 of 2 comparisons | 912/2 out of 100, in 2 of 2 comparisons | 781/1 out of 100, in 1 of 1 comparisons | 831/1 out of 100, in 1 of 1 comparisons |
| uzu | 1002/2 out of 100, in 2 of 2 comparisons | Not measured on this device. | Not measured on this device. | 1001/1 out of 100, in 1 of 1 comparisons |
| Magnitude | 842/2 out of 100, in 2 of 2 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| Ollama | 842/2 out of 100, in 2 of 2 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| LiteRT-LM | Not counted Its output check failed: 10/12 of the checks the model passes at BF16; needs 11 and the mandatory ones (needle-start, needle-end, continuation-128). Left out, failed by every BF16 configuration: none. | Not counted Its output check failed: 9/12 of the checks the model passes at BF16; needs 11 and the mandatory ones (needle-start, needle-end, continuation-128). Left out, failed by every BF16 configuration: none. | Not counted Its output check failed: 9/12 of the checks the model passes at BF16; needs 11 and the mandatory ones (needle-start, needle-end, continuation-128). Left out, failed by every BF16 configuration: none. | Not measured on this device. |
| MLX LM | Not counted Its output check failed: 10/12 of the checks the model passes at BF16; needs 11 and the mandatory ones (needle-start, needle-end, continuation-128). Left out, failed by every BF16 configuration: none. | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| Rapid-MLX | Not counted Its output check failed: 10/12 of the checks the model passes at BF16; needs 11 and the mandatory ones (needle-start, needle-end, continuation-128). Left out, failed by every BF16 configuration: none. | Not measured on this device. | Not measured on this device. | Not measured on this device. |
Speed
| Rank | Engine | Decode and TTFT |
|---|---|---|
| 1 | uzu†, approximate comparison, see details | 45.4 tok/s 1,073 ms TTFT |
| 2 | ExecuTorch†, approximate comparison, see details | 42.5 tok/s 2,033 ms TTFT |
| 3 | llama.cpp†, approximate comparison, see details | 37.6 tok/s 987 ms TTFT |
details
LFM2.5 1.2B Instruct on iPhone 12 · 4 GB · 4-bit · Archive 256 input / 128 output · iOS 27
Medians with the model warmed; model loading is excluded. Each table uses one matched comparison.
uzu: † Approximate comparison. The experiments use different effective KV cache precision: fp32, f16, uzu default. The experiments here use different 4-bit schemes: 8da4w, Q4_K_M, mirai-m.
ExecuTorch: † Approximate comparison. The experiments use different effective KV cache precision: fp32, f16, uzu default. The experiments here use different 4-bit schemes: 8da4w, Q4_K_M, mirai-m.
llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: fp32, f16, uzu default. The experiments here use different 4-bit schemes: 8da4w, Q4_K_M, mirai-m.
Memory usage by engine
| Rank | Engine | Memory |
|---|---|---|
| Not ranked | ExecuTorch | Not measured |
| Not ranked | llama.cpp | Not measured |
| Not ranked | uzu | Not measured |
details
LFM2.5 1.2B Instruct on iPhone 12 · 4 GB · 4-bit · Archive 256 input / 128 output · iOS 27
Memory ranks require comparable, calibrated measurements. Unavailable measurements remain unranked.
ExecuTorch: No peak memory measured. † Approximate comparison. The experiments use different effective KV cache precision: fp32, f16, uzu default. The experiments here use different 4-bit schemes: 8da4w, Q4_K_M, mirai-m.
llama.cpp: No peak memory measured. † Approximate comparison. The experiments use different effective KV cache precision: fp32, f16, uzu default. The experiments here use different 4-bit schemes: 8da4w, Q4_K_M, mirai-m.
uzu: No peak memory measured. † Approximate comparison. The experiments use different effective KV cache precision: fp32, f16, uzu default. The experiments here use different 4-bit schemes: 8da4w, Q4_K_M, mirai-m.
Every comparison
| Comparison | ExecuTorch | llama.cpp | uzu | LiteRT-LM |
|---|---|---|---|---|
| LFM2.5 1.2B Instruct on iPhone 12 · 4 GB 4-bit · Archive 256 input / 128 output · iOS 27 | 94†, approximate | 83†, approximate | 100†, approximate | Not run |
| LFM2.5 1.2B Instruct on OnePlus 13 · 16 GB 4-bit · Archive 2048 input / 128 output · Android 16 | 100†, approximate | 99†, approximate | Not run | Not counted |
| LFM2.5 1.2B Instruct on OnePlus 13 · 16 GB 4-bit · Archive 256 input / 128 output · Android 16 | 100†, approximate | 83†, approximate | Not run | Not counted |
| LFM2.5 1.2B Instruct on Pixel 8 · 8 GB 4-bit · Archive 256 input / 128 output · Android 16 | 100†, approximate | 78†, approximate | Not run | Not counted |
| LFM2.5 1.2B Instruct on iPhone 12 · 4 GB 4-bit · Archive 2048 input / 128 output · iOS 27 Not counted: fewer than two scored engines | Not counted | Measured†, approximate | Not counted | Not run |
| LFM2.5 1.2B Instruct on Pixel 8 · 8 GB 4-bit · Archive 2048 input / 128 output · Android 16 Not counted: fewer than two scored engines | Not counted | Not counted | Not run | Not counted |
Each engine’s Speed score in each matched comparison, out of 100, or why it has none. How rankings work