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 | MLX LM†, an approximate comparison, see details | 100 /100 Score |
| ||
| 2 | uzu†, an approximate comparison, see details | 95 /100 Score |
| ||
| 3 | Magnitude†, an approximate comparison, see details | 82 /100 Score |
| ||
| 4 | Ollama†, an approximate comparison, see details | 76 /100 Score |
| ||
| 5 | llama.cpp†, an approximate comparison, see details | 72 /100 Score |
| ||
| Not ranked in this view | ||
| Not ranked | ExecuTorch | Too few results |
| Not ranked | Rapid-MLX | Too few results |
| Not ranked | LiteRT-LM | No rival yet |
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.
† MLX LM: approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mlx-affine-4bit-g64, where most use Q4_K_M. The experiments use different effective KV cache precision: f16, model-dtype.
† uzu: approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mirai-m, where most use Q4_K_M.
† Magnitude: approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, 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: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. The experiments use different effective KV cache precision: f16, model-dtype.
† llama.cpp: approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. The experiments use different effective KV cache precision: f16, model-dtype.
ExecuTorch: Results in 1 of 3 comparisons; a rank needs 2.
Rapid-MLX: Results in 1 of 3 comparisons; a rank needs 2.
LiteRT-LM: No other engine has a result in the same comparison yet.
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 | 723/3 out of 100, in 3 of 3 comparisons | 831/1 out of 100, in 1 of 1 comparisons | 781/1 out of 100, in 1 of 1 comparisons | 602/2 out of 100, in 2 of 2 comparisons |
| ExecuTorch | 1/3Not ranked. Results in 1 of 3 comparisons; a rank needs 2. | 1001/1 out of 100, in 1 of 1 comparisons | 1001/1 out of 100, in 1 of 1 comparisons | 1/2Not ranked. Results in 1 of 2 comparisons; a rank needs 2. |
| uzu | 952/3 out of 100, in 2 of 3 comparisons | Not measured on this device. | Not measured on this device. | 1002/2 out of 100, in 2 of 2 comparisons |
| Magnitude | 822/3 out of 100, in 2 of 3 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| MLX LM | 1002/3 out of 100, in 2 of 3 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| Ollama | 763/3 out of 100, in 3 of 3 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| LiteRT-LM | No rival yet No other engine has a result in the same comparison yet. | No rival yet No other engine has a result in the same comparison yet. | No rival yet No other engine has a result in the same comparison yet. | Not measured on this device. |
| Rapid-MLX | 1/3Not ranked. Results in 1 of 3 comparisons; a rank needs 2. | Not measured on this device. | Not measured on this device. | Not measured on this device. |
Speed
| Rank | Engine | Prefill |
|---|---|---|
| 1 | uzu†, approximate comparison, see details | 13,205 tok/s |
| 2 | Magnitude†, approximate comparison, see details | 9,627 tok/s |
| 3 | Ollama†, approximate comparison, see details | 8,803 tok/s |
| 4 | llama.cpp†, approximate comparison, see details | 7,671 tok/s |
| 5 | ExecuTorch†, approximate comparison, see details | 4,923 tok/s |
| Not ranked | LiteRT-LM | Not counted |
| Not ranked | MLX LM | Not counted |
| Not ranked | Rapid-MLX | Not counted |
details
LFM2.5 1.2B Instruct on M5 Max · 64 GiB · 4-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: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mirai-m, where most use Q4_K_M.
Magnitude: † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, 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: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default.
llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default.
ExecuTorch: † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mlx-4w-g64, where most use Q4_K_M.
LiteRT-LM: 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. † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: litert-int4, where most use Q4_K_M.
MLX LM: 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. † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mlx-affine-4bit-g64, where most use Q4_K_M.
Rapid-MLX: 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. † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mlx-affine-4bit-g64, where most use Q4_K_M.
Memory usage by engine
| Rank | Engine | Memory |
|---|---|---|
| 1 | uzu†, approximate comparison, see details | 0.90 GiB |
| 2 | Ollama†, approximate comparison, see details | 0.93 GiB |
| 3 | llama.cpp†, approximate comparison, see details | 0.93 GiB |
| 4 | Magnitude†, approximate comparison, see details | 1.56 GiB |
| Not ranked | ExecuTorch | Not measured |
| Not ranked | LiteRT-LM | Not counted |
| Not ranked | MLX LM | Not counted |
| Not ranked | Rapid-MLX | Not counted |
details
LFM2.5 1.2B Instruct on M5 Max · 64 GiB · 4-bit · Archive 256 input / 128 output · macOS 27
Memory ranks require comparable, calibrated measurements. Unavailable measurements remain unranked.
uzu: † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mirai-m, where most use Q4_K_M.
Ollama: † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default.
llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default.
Magnitude: † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, 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.
ExecuTorch: No peak memory measured. † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mlx-4w-g64, where most use Q4_K_M.
LiteRT-LM: 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. † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: litert-int4, where most use Q4_K_M.
MLX LM: 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. † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mlx-affine-4bit-g64, where most use Q4_K_M.
Rapid-MLX: 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. † Approximate comparison. The experiments use different effective KV cache precision: bf16, fp32, litert-lm default, f16, magnitude default, model-dtype, uzu default. Its weights are 4-bit in another scheme: mlx-affine-4bit-g64, where most use Q4_K_M.
Every comparison
| Comparison | MLX LM | uzu | Magnitude | Ollama | llama.cpp | ExecuTorch | Rapid-MLX | LiteRT-LM |
|---|---|---|---|---|---|---|---|---|
| LFM2.5 1.2B Instruct on M5 Max · 64 GiB 4-bit · Archive 256 input / 128 output · macOS 27 | Not counted | 100†, approximate | 82†, approximate | 84†, approximate | 81†, approximate | 87†, approximate | Not counted | Not counted |
| LFM2.5 350M on M5 Max · 64 GiB 4-bit · Archive 256 input / 128 output · macOS 27 | 100†, approximate | 90†, approximate | 82†, approximate | 67†, approximate | 62†, approximate | Not counted | 100†, approximate | Not counted |
| Qwen3.8 27B on M5 Max · 64 GiB 4-bit · Archive 256 input / 128 output · macOS 27 | 100†, approximate | Unsupported | Unsupported | 78†, approximate | 75†, approximate | Not run | Not run | Not run |
| LFM2.5 2.6B on M5 Max · 64 GiB 4-bit · Archive 256 input / 128 output · macOS 27 Not counted: fewer than two scored engines | Not counted | Unsupported | Unsupported | Not counted | Not counted | Not run | Not counted | Measured†, approximate |
Each engine’s Speed score in each matched comparison, out of 100, or why it has none. How rankings work