Rankings

Explore benchmarks

Filter by model, hardware, workload and quantization. Every score links to the comparisons and experiments behind it.

Clear filters

Top Engines

Overall needs mobile results; each platform is ranked on its own until then.

Desktop

Top Engines, Desktop, Speed score.
#EngineSpeed score out of 100
1Ollama, an approximate comparison, see details Ollama 100 /100 Score
2MLX LM, an approximate comparison, see details Apple (ml-explore) 97 /100 Score
3llama.cpp, an approximate comparison, see details ggml-org 96 /100 Score
Not ranked in this view
Not rankedMagnitude Magnitude Unsupported
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

† Ollama: approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype.

† MLX LM: approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype. Its weights are 2-bit in another scheme: mlx-affine-2bit-g128, where most use Q2_0.

† llama.cpp: approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype.

Magnitude: Unsupported in 1 comparison. Magnitude 0.2.17 reports this GGUF as Unsupported (magnitude catalog inspect)

Coverage

Speed score per device, out of 100: each device ranked on its own.
Desktop
M5 Max 64 GiB macOS 27 Apple M5 Max
llama.cpp96 out of 100, in 1 of 1 comparisons
MLX LM97 out of 100, in 1 of 1 comparisons
Ollama100 out of 100, in 1 of 1 comparisons
MagnitudeUnsupported Unsupported in 1 comparison. Magnitude 0.2.17 reports this GGUF as Unsupported (magnitude catalog inspect)

Speed

TTFT ms

Generation tok/s

Speed, Ternary Bonsai 2 27B on M5 Max · 64 GiB, 2-bit · Archive 2048 input / 128 output · macOS 27
RankEngineDecode and TTFT
1Ollama, approximate comparison, see details Ollama44.6 tok/s 2,672 ms TTFT
2MLX LM, approximate comparison, see details Apple (ml-explore)43.4 tok/s 3,291 ms TTFT
3llama.cpp, approximate comparison, see details ggml-org42.7 tok/s 2,814 ms TTFT
Not rankedMagnitude MagnitudeUnsupported
details

Ternary Bonsai 2 27B on M5 Max · 64 GiB · 2-bit · Archive 2048 input / 128 output · macOS 27

Medians with the model warmed; model loading is excluded. Each table uses one matched comparison.

Ollama: † Approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype.

MLX LM: † Approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype. Its weights are 2-bit in another scheme: mlx-affine-2bit-g128, where most use Q2_0.

llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype.

Magnitude: Magnitude 0.2.17 reports this GGUF as Unsupported (magnitude catalog inspect)

Memory usage by engine

Memory usage by engine, Ternary Bonsai 2 27B on M5 Max · 64 GiB, 2-bit · Archive 2048 input / 128 output · macOS 27
RankEngineMemory
1Ollama, approximate comparison, see details Ollama8.12 GiB
2llama.cpp, approximate comparison, see details ggml-org8.12 GiB
3MLX LM, approximate comparison, see details Apple (ml-explore)14.45 GiB
Not rankedMagnitude MagnitudeUnsupported
details

Ternary Bonsai 2 27B on M5 Max · 64 GiB · 2-bit · 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, model-dtype.

llama.cpp: † Approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype.

MLX LM: † Approximate comparison. The experiments use different effective KV cache precision: f16, model-dtype. Its weights are 2-bit in another scheme: mlx-affine-2bit-g128, where most use Q2_0.

Magnitude: Magnitude 0.2.17 reports this GGUF as Unsupported (magnitude catalog inspect)

Every comparison

Comparisons in this view: each engine's score in each comparison, out of 100, or why it has none.
Comparisonllama.cppMagnitudeMLX LMOllama
Ternary Bonsai 2 27B on M5 Max · 64 GiB 2-bit · Archive 2048 input / 128 output · macOS 27 96, approximateUnsupported97, approximate100, approximate

Each engine’s Speed score in each matched comparison, out of 100, or why it has none. How rankings work