Yes — Phi-5 Mini (4B) runs at 68 tok/s on M2 with 8 GB RAM using Q4_K_M quantization via Ollama. First token latency is 0.4s. A capable open-source LLM with 4B parameters.
LLMCheck measured Phi-5 Mini on M2 using the standard methodology: Q4_K_M quantization, 256-token input, 512-token output, 3 runs averaged on a freshly-booted system.
| Metric | Value |
|---|---|
| Tokens per second | 68 tok/s |
| Time to first token | 0.4s |
| Quantization | Q4_K_M |
| Minimum RAM | 8 GB |
| Recommended engine | Ollama |
| Parameters | 4B |
| Benchmark date | 2026-05 |
Q4_K_M 4B Ollama M2
The recommended engine for Phi-5 Mini on M2 is Ollama. Install Ollama, then pull the model:
Ollama handles quantization automatically — it will download the Q4_K_M variant (~8 GB) and start an interactive chat session.
| Chip | Speed | First Token | Min RAM | Engine |
|---|---|---|---|---|
| M5 Max | 145 tok/s | 0.2s | 128 GB | MLX |
| M4 Pro | 110 tok/s | 0.3s | 24 GB | Ollama |
| M5 Pro | 95 tok/s | 0.3s | 24 GB | MLX |
| M3 | 88 tok/s | 0.3s | 16 GB | MLX |
| M1 | 50 tok/s | 0.5s | 8 GB | Ollama |
To run Phi-5 Mini on M2 you need:
See how Phi-5 Mini stacks up against other models on your specific Mac hardware.
Open Compare Tool Full Leaderboard