Yes — Qwen 4 Preview 32B-A3B (32B) runs at 60 tok/s on M4 Max with 128 GB RAM using Q4_K_M quantization via Ollama. First token latency is 0.7s. A capable open-source LLM with 32B parameters.
LLMCheck measured Qwen 4 Preview 32B-A3B on M4 Max 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 | 60 tok/s |
| Time to first token | 0.7s |
| Quantization | Q4_K_M |
| Minimum RAM | 128 GB |
| Recommended engine | Ollama |
| Parameters | 32B |
| Benchmark date | 2026-05 |
Q4_K_M 32B Ollama M4 Max
The recommended engine for Qwen 4 Preview 32B-A3B on M4 Max is Ollama. Install Ollama, then pull the model:
Ollama handles quantization automatically — it will download the Q4_K_M variant (~128 GB) and start an interactive chat session.
| Chip | Speed | First Token | Min RAM | Engine |
|---|---|---|---|---|
| M5 Max | 78 tok/s | 0.5s | 128 GB | MLX |
| M5 Max | 65 tok/s | 0.6s | 64 GB | Ollama |
| M4 Pro | 58 tok/s | 0.8s | 24 GB | Ollama |
| M5 Pro | 52 tok/s | 0.7s | 64 GB | MLX |
| M3 Max | 45 tok/s | 0.9s | 64 GB | Ollama |
| M3 Pro | 32 tok/s | 1.2s | 18 GB | Ollama |
To run Qwen 4 Preview 32B-A3B on M4 Max you need:
See how Qwen 4 Preview 32B-A3B stacks up against other models on your specific Mac hardware.
Open Compare Tool Full Leaderboard