Explore benchmarks
Filter by model, hardware, workload and quantization. Every score links to the comparisons and experiments behind it.
Top Engines
| # | Engine | Speed score out of 100 |
|---|---|---|
| 1 | uzu†, an approximate comparison, see details | 100 /100 Score |
| 2 | Rapid-MLX†, an approximate comparison, see details | 97 /100 Score |
| 3 | MLX LM†, an approximate comparison, see details | 96 /100 Score |
| 4 | Magnitude†, an approximate comparison, see details | 88 /100 Score |
| 5 | Ollama†, an approximate comparison, see details | 87 /100 Score |
| 6 | llama.cpp†, an approximate comparison, see details | 87 /100 Score |
| 7 | LiteRT-LM†, an approximate comparison, see details | 32 /100 Score |
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
† uzu: approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mirai-l, where most use Q8_0.
† Rapid-MLX: approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mlx-affine-8bit-g64, where most use Q8_0.
† MLX LM: approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mlx-affine-8bit-g64, where most use Q8_0.
† Magnitude: approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its phases were timed by the harness clock around the engine API, so they include transport that engine counters leave out.
† Ollama: approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default.
† llama.cpp: approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default.
† LiteRT-LM: approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: litert-int8, where most use Q8_0.
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 | |
| llama.cpp | 871/1 out of 100, in 1 of 1 comparisons | 1001/1 out of 100, in 1 of 1 comparisons | 871/1 out of 100, in 1 of 1 comparisons | 941/1 out of 100, in 1 of 1 comparisons |
| LiteRT-LM | 321/1 out of 100, in 1 of 1 comparisons | 861/1 out of 100, in 1 of 1 comparisons | 1001/1 out of 100, in 1 of 1 comparisons | Not measured on this device. |
| uzu | 1001/1 out of 100, in 1 of 1 comparisons | Not measured on this device. | Not measured on this device. | 1001/1 out of 100, in 1 of 1 comparisons |
| Magnitude | 881/1 out of 100, in 1 of 1 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| MLX LM | 961/1 out of 100, in 1 of 1 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| Ollama | 871/1 out of 100, in 1 of 1 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| Rapid-MLX | 971/1 out of 100, in 1 of 1 comparisons | 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 | 374.3 tok/s 75.1 ms TTFT |
| 2 | Rapid-MLX†, approximate comparison, see details | 361.9 tok/s 33.5 ms TTFT |
| 3 | MLX LM†, approximate comparison, see details | 361.1 tok/s 30.7 ms TTFT |
| 4 | Magnitude†, approximate comparison, see details | 330.1 tok/s 25.5 ms TTFT |
| 5 | Ollama†, approximate comparison, see details | 326.2 tok/s 30.5 ms TTFT |
| 6 | llama.cpp†, approximate comparison, see details | 325.0 tok/s 34.5 ms TTFT |
| 7 | LiteRT-LM†, approximate comparison, see details | 119.1 tok/s 277 ms TTFT |
details
LFM2.5 1.2B Instruct on M5 Max · 64 GiB · 8-bit · Archive 256 input / 128 output · macOS 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: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mirai-l, where most use Q8_0.
Rapid-MLX: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mlx-affine-8bit-g64, where most use Q8_0.
MLX LM: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mlx-affine-8bit-g64, where most use Q8_0.
Magnitude: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its phases were timed by the harness clock around the engine API, so they include transport that engine counters leave out.
Ollama: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default.
llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default.
LiteRT-LM: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: litert-int8, where most use Q8_0.
Memory usage by engine
| Rank | Engine | Memory |
|---|---|---|
| 1 | Ollama†, approximate comparison, see details | 1.41 GiB |
| 2 | llama.cpp†, approximate comparison, see details | 1.41 GiB |
| 3 | MLX LM†, approximate comparison, see details | 1.44 GiB |
| 4 | uzu†, approximate comparison, see details | 1.46 GiB |
| 5 | Rapid-MLX†, approximate comparison, see details | 1.69 GiB |
| 6 | LiteRT-LM†, approximate comparison, see details | 1.73 GiB |
| 7 | Magnitude†, approximate comparison, see details | 1.90 GiB |
details
LFM2.5 1.2B Instruct on M5 Max · 64 GiB · 8-bit · Archive 256 input / 128 output · macOS 27
Memory ranks require comparable, calibrated measurements. Unavailable measurements remain unranked.
Ollama: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default.
llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default.
MLX LM: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mlx-affine-8bit-g64, where most use Q8_0.
uzu: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mirai-l, where most use Q8_0.
Rapid-MLX: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: mlx-affine-8bit-g64, where most use Q8_0.
LiteRT-LM: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 8-bit in another scheme: litert-int8, where most use Q8_0.
Magnitude: † Approximate comparison. The experiments use different effective KV cache precision: litert-lm default, f16, magnitude default, model-dtype, uzu default. Its phases were timed by the harness clock around the engine API, so they include transport that engine counters leave out.
Every comparison
| Comparison | uzu | Rapid-MLX | MLX LM | Magnitude | Ollama | llama.cpp | LiteRT-LM |
|---|---|---|---|---|---|---|---|
| LFM2.5 1.2B Instruct on M5 Max · 64 GiB 8-bit · Archive 256 input / 128 output · macOS 27 | 100†, approximate | 97†, approximate | 96†, approximate | 88†, approximate | 87†, approximate | 87†, approximate | 32†, approximate |
Each engine’s Speed score in each matched comparison, out of 100, or why it has none. How rankings work