Question 1

Can I run Qwen3 Coder Next 80B A3B on a Apple M4 Ultra?

Accepted Answer

Yes — Apple M4 Ultra runs Qwen3 Coder Next 80B A3B fully on GPU at the FP16 quantization. Full GPU inference — 256 GB VRAM meets the 208 GB recommendation.

Question 2

What quantization of Qwen3 Coder Next 80B A3B should I use on a Apple M4 Ultra?

Accepted Answer

For 256 GB VRAM on the Apple M4 Ultra, the FP16 variant is the best fit. Estimated ~6.8 tokens/sec on the FP16 quantization.

Question 3

How fast does Qwen3 Coder Next 80B A3B run on Apple M4 Ultra?

Accepted Answer

Roughly 6.8 tokens/sec for FP16. Real speed depends on context length, backend (Ollama, llama.cpp, LM Studio), and KV cache size.

Question 4

What if Apple M4 Ultra is not enough for Qwen3 Coder Next 80B A3B?

Accepted Answer

Try a smaller quantization variant or run on CPU + RAM with the trade-off of slower generation.

Quantization	File Size	Min VRAM	Rec VRAM	Context	Verdict	Estimated tok/s
Q4_K_M	40 GB	46 GB	52 GB	8K / 8K	Full GPU	~21.8
Q5_K_M	50 GB	57.5 GB	65 GB	8K / 8K	Full GPU	~19
Q8_0	80 GB	92 GB	104 GB	8K / 8K	Full GPU	~13
FP16Best fit	160 GB	184 GB	208 GB	8K / 8K	Full GPU	~6.8

Can I Run Qwen3 Coder Next 80B A3B on Apple M4 Ultra?

Share this matchup

Every Qwen3 Coder Next 80B A3B quantization on Apple M4 Ultra