HARDWARE BENCHMARK & VRAM ENVELOPE • APPLE SILICON13 GB VRAM • 120 GB/s
What Models Can Run on Apple Mac Mini M4 (16GB Unified Memory)?
13 GB VRAM120 GB/s Bandwidth25W TDPMSRP: $599Used / eBay: ~$520Cloud: $0.25/hr on RunPod
Apple Mac Mini M4 (16GB Unified Memory) provides 13 GB of usable memory with 120 GB/s memory bandwidth at 25W TDP. Silent 25W desktop running 7B–8B models at full speed with 32k context.
✕ OOM25.22 GB (180%)0 t/s
Qwen 2.5 Coder 32B Instruct on Apple Mac Mini M4 (16GB Unified Memory)
✕ Out of Memory — Exceeds VRAM Capacity
✕ OOM
Required VRAM
25.22 GB
180% of hardware capacity (14 GB)Est. Decode Speed
0 tok/s
Dense model @ 120 GB/sAvailable Headroom
-11.22 GB
Needs +11.22 GB more VRAMMemory Allocation & Headroom
25.22 GB allocated of 14 GB capacity
180% utilized
Weights: 19.5 GB
KV: (32K): 4.29 GB
Overhead: 1.43 GB
Deficit: -11.22 GB
Architecture & Throughput Transparency: Single-user decode speed (~0 tok/s) is memory-bandwidth bound at batch size = 1 on Apple Mac Mini M4 (16GB Unified Memory) (120 GB/s). Prompt prefill / TTFT (~0 tok/s) is compute-bound and utilizes GPU Tensor cores. Actual latency varies with FlashAttention-2 and runtime engine (Ollama/vLLM/llama.cpp).
Out of VRAM? Don't downgrade model quality.
Out of VRAMSpin up an on-demand cloud GPU instance in 60 seconds with pre-installed Ollama and vLLM. (Qwen 2.5 Coder 32B Instruct @ 25.22 GB VRAM)
RunPod Cloud GPUsBest 1-Click Ollama & vLLM
$0.38/hr (RTX 4090 Spot) – $2.39/hr (H100)Vast.ai GPU MarketplaceCheapest Dual RTX 3090 / 4090
$0.18/hr (RTX 3060) – $0.78/hr (2x RTX 3090)Lambda GPU CloudEnterprise 8x H100 Clusters
$2.49/hr (H100) – $19.92/hr (8x H100)Quantization Precision Matrix
32K ctx • KV: Q8_0| Quantization | Weights | KV Cache | Total VRAM | Est. Speed | Fit Status |
|---|---|---|---|---|---|
| FP1616 bpw | 65 GB | 4.29 GB | 72.32 GB (517%) | 0 t/s | OOM |
| Q8_08.5 bpw | 34.45 GB | 4.29 GB | 40.7 GB (291%) | 0 t/s | OOM |
| Q6_K6.56 bpw | 26.65 GB | 4.29 GB | 32.62 GB (233%) | 0 t/s | OOM |
| Q5_K_M5.68 bpw | 23.07 GB | 4.29 GB | 28.92 GB (207%) | 0 t/s | OOM |
| Q4_K_M4.8 bpw | 19.5 GB | 4.29 GB | 25.22 GB (180%) | 0 t/s | OOM |
| FP44.25 bpw | 17.23 GB | 4.29 GB | 22.87 GB (163%) | 0 t/s | OOM |
| IQ3_XS3.3 bpw | 13.32 GB | 4.29 GB | 18.83 GB (135%) | ~1 t/s | PARTIAL |
All Open-Weights Models Tested on Apple Mac Mini M4 (16GB Unified Memory) (Q4_K_M @ 32K Context)
Standard Q4_K_M precision with 32K context window and Q8_0 KV cache
| Model | Architecture | Total VRAM Needed | Headroom | Est. Decode Speed | Compatibility | Detailed Config |
|---|---|---|---|---|---|---|
| Qwen 2.5 Coder 32B Instruct Alibaba Cloud | 32.5B Dense | 25.22 GB(194%) | -12.22 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| DeepSeek-R1 (671B MoE) DeepSeek | 671B MoE | 427.26 GB(999%) | -414.26 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| DeepSeek-V3 (671B MoE) DeepSeek | 671B MoE | 427.26 GB(999%) | -414.26 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Meta Llama 3.3 70B Instruct Meta AI | 70.6B Dense | 49.96 GB(384%) | -36.96 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| DeepSeek-R1 Distill Qwen 32B DeepSeek | 32.8B Dense | 25.41 GB(195%) | -12.41 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| DeepSeek-R1 Distill Qwen 14B DeepSeek | 14.8B Dense | 13.16 GB(101%) | -0.16 GB | ~2 tok/s | ! PARTIAL | Configure CLI → |
| Qwen 2.5 72B Instruct Alibaba Cloud | 72.7B Dense | 51.27 GB(394%) | -38.27 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Meta Llama 3.1 8B Instruct Meta AI | 8B Dense | 7.87 GB(61%) | +5.13 GB | ~13 tok/s | ✓ FITS | Configure CLI → |
| Microsoft Phi-4 14B Microsoft | 14.7B Dense | 13.24 GB(102%) | -0.24 GB | ~2 tok/s | ! PARTIAL | Configure CLI → |
| DeepSeek Coder V2 236B MoE DeepSeek | 236B MoE | 147.91 GB(999%) | -134.91 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Mistral Large 2 (123B Instruct) Mistral AI | 123B Dense | 83.04 GB(639%) | -70.04 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Meta Llama 3.1 405B Instruct Meta AI | 405B Dense | 260.72 GB(999%) | -247.72 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Google Gemma 2 27B Instruct Google | 27.2B Dense | 19.18 GB(148%) | -6.18 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Google Gemma 2 9B Instruct Google | 9.2B Dense | 7.87 GB(61%) | +5.13 GB | ~12 tok/s | ✓ FITS | Configure CLI → |
| Qwen 3.8 27B Instruct Alibaba Cloud | 27.2B Dense | 21.93 GB(169%) | -8.93 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Qwen 3.8 235B-A22B MoE Alibaba Cloud | 235B MoE | 149.84 GB(999%) | -136.84 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Qwen 3.6 35B-A3B MoE Alibaba Cloud | 35B MoE | 24.1 GB(185%) | -11.1 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| DeepSeek-V4.1-Flash (552B Engram MoE) DeepSeek | 552B MoE | 343.85 GB(999%) | -330.85 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Meta Muse Spark Open (816B-A42B MoE) Meta AI | 816B MoE | 512.86 GB(999%) | -499.86 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Google Gemma 4 26B MoE Google | 26.4B MoE | 20.23 GB(156%) | -7.23 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Google Gemma 4 E4B (Edge 4B) Google | 4.1B Dense | 4.24 GB(33%) | +8.76 GB | ~27 tok/s | ✓ FITS | Configure CLI → |
| Meta Llama 4 Scout (109B-A17B MoE) Meta AI | 109B MoE | 71.66 GB(551%) | -58.66 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| Microsoft Phi-4 Mini 3.8B Microsoft | 3.8B Dense | 3.92 GB(30%) | +9.08 GB | ~29 tok/s | ✓ FITS | Configure CLI → |
| Mistral Small 3 24B Instruct Mistral AI | 24B Dense | 18.33 GB(141%) | -5.33 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| DeepSeek-R1 Distill Llama 70B DeepSeek | 70.6B Dense | 49.96 GB(384%) | -36.96 GB | 0 (OOM) | ✕ OOM | Configure CLI → |
| DeepSeek-R1 Distill Qwen 7B DeepSeek | 7.6B Dense | 6.41 GB(49%) | +6.59 GB | ~15 tok/s | ✓ FITS | Configure CLI → |
| Qwen 2.5 Coder 7B Instruct Alibaba Cloud | 7.6B Dense | 6.41 GB(49%) | +6.59 GB | ~15 tok/s | ✓ FITS | Configure CLI → |