Apple M5 (32GB)
The Apple M5 (32GB) has 32 GB VRAM and 153 GB/s memory bandwidth. It can run 43 of our 70 tracked models natively in VRAM at 8k context.
With 32 GB LPDDR5X, the Apple M5 (32GB) is a laptop-tier GPU that can run 43 models natively. It handles 70B-class models at Q4 quantization.
The Apple M5 (32GB) pairs 32GB of unified LPDDR5X memory with 153 GB/s bandwidth, making it a capable entry-level chip for local LLM inference on MacBook. Qwen 3.6 27B runs at Q4_K_M with comfortable headroom, and Gemma 4 26B MoE fits easily at Q4_K_M — both fully in-memory via MLX and llama.cpp without requiring a Pro or Max upgrade. For users who primarily run 27B-class models, the base M5 (32GB) is a cost-effective on-device AI platform.
Apple M5 (32GB): 2025 Apple Silicon with 32GB LPDDR5X at 153 GB/s. 10-core CPU (4P+6E).
7B-14B at Q4 native. 27B tight. ~8-12 t/s for 7B via MLX.
MLX framework optimized. llama.cpp Metal backend fully supported. Best laptop efficiency.
| Vendor | Apple |
| Architecture | Apple M5 |
| CPU cores | 10 (4P + 6E) |
| VRAM | 32 GB (unified) |
| Memory type | LPDDR5X |
| Memory bandwidth | 153 GB/s |
| Compute backend | METAL |
| Tier | Laptop |
| Released | 2025 |
| Models (native) | 43 / 70 |
| Models (offload) | 0 / 70 |
Software: MLX gives the best performance on Apple Silicon; llama.cpp Metal backend is a solid alternative. Both are well-supported by Ollama.
Popular models for this GPU
Models this GPU runs natively in VRAM (43)
- Mixtral 8x7B Instruct v0.146.7B · MMLU-Pro 29.7Q3_K_M · ~30.3 t/s
- Command-R 35B35B · MMLU-Pro 33.0Q2_K · ~13.3 t/s
- Qwen 3.5 35B-A3B (MoE)35B · MMLU-Pro 84.2Q5_K_M · ~87.1 t/s
- Qwen 3.6 35B35B · MMLU-Pro 85.2Q5_K_M · ~6.8 t/s
- Yi 1.5 34B Chat34.4B · MMLU-Pro 37.0Q5_K_M · ~6.9 t/s
- Qwen3 32B32.8B · MMLU-Pro 65.5Q5_K_M · ~7.2 t/s
- Qwen 2.5 32B Instruct32.5B · MMLU-Pro 69.0Q5_K_M · ~7.3 t/s
- Qwen 2.5 Coder 32B Instruct32.5B · MMLU-Pro 50.4Q5_K_M · ~7.3 t/s
- DeepSeek R1 Distill Qwen 32B32.5B · MMLU-Pro 65.0Q5_K_M · ~7.3 t/s
- Nemotron 3 Nano 30B32B · MMLU-Pro 78.3Q5_K_M · ~87.1 t/s
- Gemma 4 31B31B · MMLU-Pro 85.2Q5_K_M · ~7.7 t/s
- Qwen3 30B-A3B (MoE)30B · MMLU-Pro 61.5Q5_K_M · ~87.1 t/s
- Gemma 2 27B Instruct27.2B · MMLU-Pro 38.0Q5_K_M · ~8.7 t/s
- Gemma 3 27B Instruct27B · MMLU-Pro 67.5Q6_K · ~6.9 t/s
- Qwen 3.6 27B27B · MMLU-Pro 86.2Q6_K · ~6.9 t/s
- Gemma 4 26B (MoE)26B · MMLU-Pro 82.6Q6_K · ~54 t/s
- Mistral Small 3.1 24B Instruct24B · MMLU-Pro 66.8Q6_K · ~7.8 t/s
- Mistral Small 22B22.2B · MMLU-Pro 49.2Q8_0 · ~6.9 t/s
- GPT-OSS 20B21B · MMLU-Pro 67.9Q8_0 · ~42.1 t/s
- Qwen3 14B14.8B · MMLU-Pro 61.0Q8_0 · ~10.3 t/s
- Qwen 2.5 14B Instruct14.7B · MMLU-Pro 63.7Q8_0 · ~10.4 t/s
- Phi-4 14B Instruct14B · MMLU-Pro 70.4Q8_0 · ~10.9 t/s
- Mistral Nemo 12B Instruct12.2B · MMLU-Pro 35.6Q8_0 · ~12.5 t/s
- Gemma 3 12B Instruct12.2B · MMLU-Pro 60.6Q8_0 · ~12.5 t/s
- Gemma 2 9B Instruct9.2B · MMLU-Pro 32.0BF16 · ~8.3 t/s
- Llama 3.1 8B Instruct8B · MMLU-Pro 48.3BF16 · ~9.6 t/s
- DeepSeek R1 Distill Llama 8B8B · MMLU-Pro 41.0BF16 · ~9.6 t/s
- Qwen3 8B8B · MMLU-Pro 56.7BF16 · ~9.6 t/s
- Qwen 2.5 7B Instruct7.6B · MMLU-Pro 56.3BF16 · ~10.1 t/s
- Mistral 7B Instruct v0.37.25B · MMLU-Pro 30.0BF16 · ~10.6 t/s
- Gemma 3 4B Instruct4B · MMLU-Pro 43.6FP32 · ~9.6 t/s
- Gemma 4 E4B4B · MMLU-Pro 69.4FP32 · ~9.6 t/s
- Phi-3.5 Mini Instruct3.8B · MMLU-Pro 47.4FP32 · ~10.1 t/s
- Llama 3.2 3B Instruct3.2B · MMLU-Pro 24.0FP32 · ~12 t/s
- Qwen 2.5 3B Instruct3.1B · MMLU-Pro 32.4FP32 · ~12.3 t/s
- Gemma 2 2B Instruct2.6B · MMLU-Pro 17.8FP32 · ~14.7 t/s
- Gemma 4 E2B2B · MMLU-Pro 60.0FP32 · ~19.1 t/s
- SmolLM2 1.7B Instruct1.7B · MMLU-Pro 19.0FP32 · ~22.5 t/s
- Qwen 2.5 1.5B Instruct1.5B · MMLU-Pro 16.8FP32 · ~25.5 t/s
- Llama 3.2 1B Instruct1.24B · MMLU-Pro 12.5FP32 · ~30.8 t/s
- Gemma 3 1B Instruct1B · MMLU-Pro 14.7FP32 · ~38.3 t/s
- Qwen 2.5 0.5B Instruct0.5B · MMLU-Pro 10.0FP32 · ~76.5 t/s
- SmolLM2 360M Instruct0.36B · MMLU-Pro 8.0FP32 · ~106.3 t/s
Too large for this GPU (27)
- Llama 3.3 70B Instruct
- Qwen 2.5 72B Instruct
- DeepSeek R1 Distill Llama 70B
- Llama 3.1 70B Instruct
- Mixtral 8x22B Instruct v0.1
- Llama 3.1 405B Instruct
- DeepSeek V3 671B
- DeepSeek R1 671B
- Llama 4 Scout 109B
- Llama 4 Maverick 400B
- Qwen3 235B-A22B (MoE)
- MiniMax M1 456B
- GPT-OSS 120B
- GLM-4.5 355B
- GLM-4.5 Air 106B
- GLM-4.6 355B
- GLM-4.6V 106B
- GLM-4.7 358B
- Qwen 3.5 122B-A10B (MoE)
- MiniMax M2.5 229B
- GLM-5 744B
- MiniMax M2.7 229B
- Nemotron 3 Super 120B
- Kimi K2.6
- GLM-5.1 754B
- DeepSeek V4 Pro 1.6T
- DeepSeek V4 Flash 284B
Frequently asked questions
- How much VRAM does the Apple M5 (32GB) have?
- The Apple M5 (32GB) has 32 GB of LPDDR5X with 153 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
- What is the Apple M5 (32GB) best for?
- With 32 GB of VRAM, the Apple M5 (32GB) is well-suited for running 7B–32B models at Q4 with room for context, making it a great all-rounder for local LLM inference.
- What LLMs can the Apple M5 (32GB) run locally?
- The Apple M5 (32GB) can run 43 of the 70 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Llama 3.1 8B Instruct at BF16, Llama 3.2 3B Instruct at FP32, Llama 3.2 1B Instruct at FP32.
- Can the Apple M5 (32GB) run Llama 3.3 70B Instruct?
- The Apple M5 (32GB) does not have enough VRAM to run Llama 3.3 70B Instruct. You would need more VRAM or a lower quantization level.
- Can the Apple M5 (32GB) run Qwen 3.6 27B?
- Yes. The Apple M5 (32GB) runs Qwen 3.6 27B natively in VRAM at Q6_K quantization, achieving approximately 6.9 tokens per second.
- Can the Apple M5 (32GB) run Llama 3.1 8B Instruct?
- Yes. The Apple M5 (32GB) runs Llama 3.1 8B Instruct natively in VRAM at BF16 quantization, achieving approximately 9.6 tokens per second.