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 | Rapid-MLX†, an approximate comparison, see details | 100 /100 Score |
| 2 | MLX LM†, an approximate comparison, see details | 99 /100 Score |
| 3 | Magnitude†, an approximate comparison, see details | 96 /100 Score |
| 4 | Ollama†, an approximate comparison, see details | 88 /100 Score |
| 5 | llama.cpp†, an approximate comparison, see details | 83 /100 Score |
| 6 | ZML†, an approximate comparison, see details | 49 /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
† Rapid-MLX: approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
† MLX LM: approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
† Magnitude: approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml 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: f16, magnitude default, model-dtype, zml default.
† llama.cpp: approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
† ZML: approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default. Its phases were timed by the harness clock around the engine API, so they include transport that engine counters leave out.
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 | 831/1 out of 100, in 1 of 1 comparisons | No rival yet No other engine has a result in the same comparison yet. | Failed Failed in 1 comparison. No trial was valid. | No rival yet No other engine has a result in the same comparison yet. |
| Magnitude | 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. |
| MLX LM | 991/1 out of 100, in 1 of 1 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| Ollama | 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. |
| Rapid-MLX | 1001/1 out of 100, in 1 of 1 comparisons | Not measured on this device. | Not measured on this device. | Not measured on this device. |
| ZML | 491/1 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 | Rapid-MLX†, approximate comparison, see details | 532.2 tok/s 46.9 ms TTFT |
| 2 | MLX LM†, approximate comparison, see details | 526.9 tok/s 38.7 ms TTFT |
| 3 | Magnitude†, approximate comparison, see details | 510.1 tok/s 48.9 ms TTFT |
| 4 | Ollama†, approximate comparison, see details | 469.0 tok/s 63.4 ms TTFT |
| 5 | llama.cpp†, approximate comparison, see details | 442.7 tok/s 66.2 ms TTFT |
| 6 | ZML†, approximate comparison, see details | 261.0 tok/s 59.6 ms TTFT |
details
LFM2.5 350M on M5 Max · 64 GiB · bf16 · Archive 2048 input / 128 output · macOS 27
Medians with the model warmed; model loading is excluded. Each table uses one matched comparison.
Rapid-MLX: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
MLX LM: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
Magnitude: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml 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: f16, magnitude default, model-dtype, zml default.
llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
ZML: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default. Its phases were timed by the harness clock around the engine API, so they include transport that engine counters leave out.
Memory usage by engine
| Rank | Engine | Memory |
|---|---|---|
| 1 | Ollama†, approximate comparison, see details | 0.92 GiB |
| 2 | llama.cpp†, approximate comparison, see details | 0.92 GiB |
| 3 | MLX LM†, approximate comparison, see details | 1.01 GiB |
| 4 | Rapid-MLX†, approximate comparison, see details | 1.28 GiB |
| 5 | Magnitude†, approximate comparison, see details | 1.35 GiB |
| 6 | ZML†, approximate comparison, see details | 4.87 GiB |
details
LFM2.5 350M on M5 Max · 64 GiB · bf16 · Archive 2048 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: f16, magnitude default, model-dtype, zml default.
llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
MLX LM: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
Rapid-MLX: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default.
Magnitude: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml default. Its phases were timed by the harness clock around the engine API, so they include transport that engine counters leave out.
ZML: † Approximate comparison. The experiments use different effective KV cache precision: f16, magnitude default, model-dtype, zml 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 | Rapid-MLX | MLX LM | Magnitude | Ollama | llama.cpp | ZML |
|---|---|---|---|---|---|---|
| LFM2.5 350M on M5 Max · 64 GiB bf16 · Archive 2048 input / 128 output · macOS 27 | 100†, approximate | 99†, approximate | 96†, approximate | 88†, approximate | 83†, approximate | 49†, approximate |
Each engine’s Speed score in each matched comparison, out of 100, or why it has none. How rankings work