Apple M3
The Apple M3 ships in 8–24 GB unified-memory configurations at 100 GB/s. Across those configurations it runs 39 of our 84 tracked models natively in VRAM at 8k context.
More memory means more of our tracked models fit natively — see which configuration you need below.
| Configuration | Bandwidth | CPU cores | Native models | + Offload |
|---|---|---|---|---|
| 24 GB | 100 GB/s | 8 (4P + 4E) | 39 / 84 | 0 |
| 16 GB | 100 GB/s | 8 (4P + 4E) | 26 / 84 | 0 |
| 8 GB | 100 GB/s | 8 (4P + 4E) | 0 / 84 | 0 |
How much of the Apple M3's memory is actually usable?
macOS and background apps need a slice of the pool before a model gets to use it — this site reserves 8GB on every unified-memory GPU, the same baseline used everywhere else on this site. What's left is real headroom for a model's weights and KV cache:
Apple M3 (24GB)
With 24 GB LPDDR5 at 100 GB/s, this configuration runs 39 models natively. It comfortably runs 7B–32B models at Q4; 70B-class models typically need CPU offload.
Apple M3 (24GB): the top memory tier of the base M3 chip, a BTO-only upgrade over the 8GB and 16GB standard configurations at the same 8-core CPU (4P+4E) and 100 GB/s bandwidth.
With 16 GB of real headroom, this configuration is where mixture-of-experts models start to separate from same-size dense ones: Qwen 3.5 35B-A3B — a 35B-parameter model — fits at Q2_K (15.12 GB) at 20.1 tok/s, while the dense Qwen3 32B needs the same Q2_K precision at a similar footprint (15.50 GB) but manages only 5.8 tok/s, a 3.5x gap driven entirely by MoE decode reading far fewer bytes per token. 39 of the 84 tracked models fit natively, though most 27B+ dense models are squeezed into Q2_K or Q3_K_M here — real capability, not much precision headroom.
MLX and llama.cpp's Metal backend are both mature on this chip. This configuration's largest fits (~15-16 GB) stay comfortably under the default ~18 GB Metal working-set limit (75% of 24 GB), so the manual wired_limit override isn't necessary here.
Models the 24 GB configuration runs natively (39)
- Qwen 3.5 35B-A3B (MoE)35B · MMLU-Pro 85.3Q2_K · ~20.1 t/s
- Qwen3 32B32.8B · MMLU-Pro 65.5Q2_K · ~5.8 t/s
- Nemotron 3 Nano 30B32B · MMLU-Pro 78.3Q2_K · ~18.8 t/s
- Qwen3 30B-A3B (MoE)30B · MMLU-Pro 61.5Q2_K · ~17.3 t/s
- Gemma 2 27B Instruct27.2B · MMLU-Pro 38.0Q2_K · ~5.9 t/s
Show 34 more
- Gemma 3 27B Instruct27B · MMLU-Pro 67.5Q2_K · ~6.8 t/s
- Qwen 3.6 27B27B · MMLU-Pro 86.2Q3_K_M · ~5.9 t/s
- Bonsai 27B27B · MMLU-Pro 81.5Ternary (Q2_0) · ~9.1 t/s
- Gemma 4 26B (MoE)26B · MMLU-Pro 82.6Q3_K_M · ~12.4 t/s
- Mistral Small 3.1 24B Instruct24B · MMLU-Pro 66.8Q3_K_M · ~6.2 t/s
- Mistral Small 22B22.2B · MMLU-Pro 49.2Q3_K_M · ~6.4 t/s
- GPT-OSS 20B21B · MMLU-Pro 67.9Q4_K_M · ~10.6 t/s
- Qwen3 14B14.8B · MMLU-Pro 61.0Q6_K · ~5.9 t/s
- Qwen 2.5 14B Instruct14.7B · MMLU-Pro 63.7Q6_K · ~5.8 t/s
- Phi-4 14B Instruct14B · MMLU-Pro 70.4Q6_K · ~6.2 t/s
- Mistral Nemo 12B Instruct12.2B · MMLU-Pro 35.6Q6_K · ~7 t/s
- Gemma 3 12B Instruct12.2B · MMLU-Pro 60.6Q8_0 · ~5.7 t/s
- Gemma 4 12B (Unified)12B · MMLU-Pro 77.2Q6_K · ~6.1 t/s
- Gemma 2 9B Instruct9.2B · MMLU-Pro 32.0Q8_0 · ~6.4 t/s
- Qwen 3.5 9B9B · MMLU-Pro 82.5Q8_0 · ~8.1 t/s
- Llama 3.1 8B Instruct8B · MMLU-Pro 48.3Q8_0 · ~8.4 t/s
- DeepSeek R1 Distill Llama 8B8B · MMLU-Pro 41.0Q8_0 · ~8.4 t/s
- Qwen3 8B8B · MMLU-Pro 56.7Q8_0 · ~8.2 t/s
- Qwen 2.5 7B Instruct7.6B · MMLU-Pro 56.3Q8_0 · ~9.4 t/s
- Mistral 7B Instruct v0.37.25B · MMLU-Pro 30.0Q8_0 · ~9.1 t/s
- Gemma 3 4B Instruct4B · MMLU-Pro 43.6BF16 · ~9.4 t/s
- Gemma 4 E4B4B · MMLU-Pro 69.4BF16 · ~8.9 t/s
- Phi-3.5 Mini Instruct3.8B · MMLU-Pro 47.4BF16 · ~7.4 t/s
- Phi-4-mini Instruct3.8B · MMLU-Pro 67.3BF16 · ~9.2 t/s
- Llama 3.2 3B Instruct3.2B · MMLU-Pro 24.0FP32 · ~5.8 t/s
- Qwen 2.5 3B Instruct3.1B · MMLU-Pro 32.4FP32 · ~6.3 t/s
- Gemma 2 2B Instruct2.6B · MMLU-Pro 17.8FP32 · ~7.1 t/s
- Gemma 4 E2B2B · MMLU-Pro 60.0FP32 · ~9.5 t/s
- SmolLM2 1.7B Instruct1.7B · MMLU-Pro 19.0FP32 · ~9.5 t/s
- Qwen 2.5 1.5B Instruct1.5B · MMLU-Pro 16.8FP32 · ~12.8 t/s
- Llama 3.2 1B Instruct1.24B · MMLU-Pro 12.5FP32 · ~15.3 t/s
- Gemma 3 1B Instruct1B · MMLU-Pro 14.7FP32 · ~18.5 t/s
- Qwen 2.5 0.5B Instruct0.5B · MMLU-Pro 10.0FP32 · ~38.1 t/s
- SmolLM2 360M Instruct0.36B · MMLU-Pro 8.0FP32 · ~45.1 t/s
Apple M3 (16GB)
With 16 GB LPDDR5 at 100 GB/s, this configuration runs 26 models natively. It handles smaller models (7B–14B) at Q4–Q5 quantization.
Apple M3 (16GB): the 16GB build of the base M3 chip, same 8-core CPU (4P+4E) and 100 GB/s bandwidth as the 8GB and 24GB configurations — only the memory pool changes. 100 GB/s is a 47% jump over the base M1's 68 GB/s, but flat against the base M2's own 100 GB/s — the only tier in the M3 generation where bandwidth didn't move at all from the chip it replaced.
With 8 GB of real headroom after the standard 8 GB reservation, this site's calculator fits Llama 3.1 8B at Q5_K_M (7.58 GB, 11.8 tok/s) and Qwen2.5 7B at Q6_K (7.51 GB, 11.9 tok/s) — about 1.5x the identically-sized base M1 16GB's speed on the same two models (8 and 8.1 tok/s there), tracking the 100-vs-68-GB/s bandwidth ratio closely. 26 of the 84 tracked models fit natively — the same count as the base M1 and M2 16GB configurations at this identical 8 GB headroom, since capacity, not bandwidth, decides the model list.
MLX and llama.cpp's Metal backend are both mature on this chip. Nothing this configuration fits comes close to the default ~12 GB Metal working-set limit (75% of 16 GB) — the largest fit here is under 8 GB — so the manual wired_limit override other Apple Silicon pages need doesn't apply at this capacity.
Models the 16 GB configuration runs natively (26)
- Bonsai 27B27B · MMLU-Pro 81.51-bit (Q1_0) · ~14.8 t/s
- Qwen3 14B14.8B · MMLU-Pro 61.0Q2_K · ~11.5 t/s
- Phi-4 14B Instruct14B · MMLU-Pro 70.4Q2_K · ~12 t/s
- Mistral Nemo 12B Instruct12.2B · MMLU-Pro 35.6Q2_K · ~13.4 t/s
- Gemma 3 12B Instruct12.2B · MMLU-Pro 60.6Q3_K_M · ~11.6 t/s
Show 21 more
- Gemma 2 9B Instruct9.2B · MMLU-Pro 32.0Q2_K · ~12.7 t/s
- Qwen 3.5 9B9B · MMLU-Pro 82.5Q5_K_M · ~12 t/s
- Llama 3.1 8B Instruct8B · MMLU-Pro 48.3Q5_K_M · ~11.8 t/s
- DeepSeek R1 Distill Llama 8B8B · MMLU-Pro 41.0Q5_K_M · ~11.8 t/s
- Qwen3 8B8B · MMLU-Pro 56.7Q5_K_M · ~11.6 t/s
- Qwen 2.5 7B Instruct7.6B · MMLU-Pro 56.3Q6_K · ~11.9 t/s
- Mistral 7B Instruct v0.37.25B · MMLU-Pro 30.0Q6_K · ~11.4 t/s
- Gemma 3 4B Instruct4B · MMLU-Pro 43.6Q8_0 · ~16.8 t/s
- Gemma 4 E4B4B · MMLU-Pro 69.4Q8_0 · ~15.2 t/s
- Phi-3.5 Mini Instruct3.8B · MMLU-Pro 47.4Q6_K · ~12.6 t/s
- Phi-4-mini Instruct3.8B · MMLU-Pro 67.3Q8_0 · ~15.6 t/s
- Llama 3.2 3B Instruct3.2B · MMLU-Pro 24.0Q8_0 · ~18.4 t/s
- Qwen 2.5 3B Instruct3.1B · MMLU-Pro 32.4BF16 · ~12.3 t/s
- Gemma 2 2B Instruct2.6B · MMLU-Pro 17.8BF16 · ~13.2 t/s
- Gemma 4 E2B2B · MMLU-Pro 60.0BF16 · ~18.2 t/s
- SmolLM2 1.7B Instruct1.7B · MMLU-Pro 19.0BF16 · ~16 t/s
- Qwen 2.5 1.5B Instruct1.5B · MMLU-Pro 16.8FP32 · ~12.8 t/s
- Llama 3.2 1B Instruct1.24B · MMLU-Pro 12.5FP32 · ~15.3 t/s
- Gemma 3 1B Instruct1B · MMLU-Pro 14.7FP32 · ~18.5 t/s
- Qwen 2.5 0.5B Instruct0.5B · MMLU-Pro 10.0FP32 · ~38.1 t/s
- SmolLM2 360M Instruct0.36B · MMLU-Pro 8.0FP32 · ~45.1 t/s
Apple M3 (8GB)
With 8 GB LPDDR5 at 100 GB/s, this configuration runs 0 models natively. It's best for smaller models under 8B parameters.
Apple M3 (8GB): the entry tier of the November 2023 M3 generation — 8-core CPU (4P+4E), 8- or 10-core GPU depending on SKU, and 100 GB/s LPDDR5 bandwidth, identical to the base M2's bandwidth two years earlier. Ships in the base 14-inch MacBook Pro, iMac, and, from March 2024, the 13-inch and 15-inch MacBook Air.
This site reserves 8 GB of unified memory for macOS and background apps on every Apple Silicon GPU — the same baseline used everywhere on this site — which leaves this 8 GB machine with 0 GB of real headroom for a model's weights and KV cache. None of the 84 models tracked on this site fit natively in memory here; even the smallest quantized models need more room than the reservation leaves free.
MLX and llama.cpp's Metal backend both run on this chip, but neither changes the arithmetic: an 8 GB pool minus an 8 GB OS reservation is 0 GB free, regardless of software. Anyone shopping specifically for local LLM inference should look at the 16GB M3 at minimum, or an M3 Pro for real headroom.
Models the 8 GB configuration runs natively (0)
None.
Too large for any Apple M3 configuration (45)
- Llama 3.3 70B Instruct
- Qwen 2.5 72B Instruct
- Qwen 2.5 32B Instruct
- Qwen 2.5 Coder 32B Instruct
- Mixtral 8x7B Instruct v0.1
- DeepSeek R1 Distill Llama 70B
- DeepSeek R1 Distill Qwen 32B
- Command-R 35B
- Yi 1.5 34B Chat
- 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
- Gemma 4 31B
- Qwen 3.5 122B-A10B (MoE)
- MiniMax M2.5 229B
- GLM-5 744B
- MiniMax M2.7 229B
- Nemotron 3 Super 120B
- Qwen 3.6 35B
- Kimi K2.6
- GLM-5.1 754B
- DeepSeek V4 Pro 1.6T
- DeepSeek V4 Flash 284B
- Mistral Medium 3.5 128B
- GLM-5.2 753B
- Nemotron 3 Ultra 550B-A55B
- Step 3.5 Flash
- Step 3.7 Flash
- MiMo V2.5 Pro
- Kimi K2.5
- MiniMax M3
- Inkling
- Kimi K3
Frequently asked questions
- How much memory does the Apple M3 have?
- The Apple M3 ships in 3 unified-memory configurations: 24 GB, 16 GB, 8 GB, all at 100 GB/s.
- Should I get the 8 GB or 24 GB Apple M3?
- The 8 GB configuration doesn't run any of the models tracked on this site natively in VRAM — after this site's standard OS-memory reservation, it has no real headroom left for a model's weights. The 24 GB configuration is the practical minimum for local LLM inference on this chip, fitting Qwen 3.5 35B-A3B (MoE), Qwen3 32B, Nemotron 3 Nano 30B, and 36 more models natively.
- How much VRAM does the Apple M3 (24GB) have?
- The Apple M3 (24GB) has 24 GB of LPDDR5 with 100 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
- What is the Apple M3 (24GB) best for?
- With 24 GB of VRAM, the Apple M3 (24GB) 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 M3 (24GB) run locally?
- The Apple M3 (24GB) can run 39 of the 84 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Llama 3.1 8B Instruct at Q8_0, Llama 3.2 3B Instruct at FP32, Llama 3.2 1B Instruct at FP32.
- Can the Apple M3 (24GB) run Llama 3.3 70B Instruct?
- The Apple M3 (24GB) does not have enough VRAM to run Llama 3.3 70B Instruct. You would need more VRAM or a lower quantization level.
Show 14 more questions
- Can the Apple M3 (24GB) run Qwen 3.6 27B?
- Yes. The Apple M3 (24GB) runs Qwen 3.6 27B natively in VRAM at Q3_K_M quantization, achieving approximately 5.9 tokens per second.
- Can the Apple M3 (24GB) run Llama 3.1 8B Instruct?
- Yes. The Apple M3 (24GB) runs Llama 3.1 8B Instruct natively in VRAM at Q8_0 quantization, achieving approximately 8.4 tokens per second.
- How much VRAM does the Apple M3 (16GB) have?
- The Apple M3 (16GB) has 16 GB of LPDDR5 with 100 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
- What is the Apple M3 (16GB) best for?
- With 16 GB of VRAM, the Apple M3 (16GB) handles smaller models (7B–14B) at Q4–Q5 quantization — ideal for entry-level local LLM experimentation and lightweight inference.
- What LLMs can the Apple M3 (16GB) run locally?
- The Apple M3 (16GB) can run 26 of the 84 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Llama 3.1 8B Instruct at Q5_K_M, Llama 3.2 3B Instruct at Q8_0, Llama 3.2 1B Instruct at FP32.
- Can the Apple M3 (16GB) run Llama 3.3 70B Instruct?
- The Apple M3 (16GB) 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 M3 (16GB) run Qwen 3.6 27B?
- The Apple M3 (16GB) does not have enough VRAM to run Qwen 3.6 27B. You would need more VRAM or a lower quantization level.
- Can the Apple M3 (16GB) run Llama 3.1 8B Instruct?
- Yes. The Apple M3 (16GB) runs Llama 3.1 8B Instruct natively in VRAM at Q5_K_M quantization, achieving approximately 11.8 tokens per second.
- How much VRAM does the Apple M3 (8GB) have?
- The Apple M3 (8GB) has 8 GB of LPDDR5 with 100 GB/s memory bandwidth (unified system memory, shared between CPU and GPU).
- What is the Apple M3 (8GB) best for?
- With 8 GB of VRAM, the Apple M3 (8GB) is best for running compact models (1B–8B) at low quantization, suitable for edge inference, prototyping, and lightweight tasks.
- What LLMs can the Apple M3 (8GB) run locally?
- The Apple M3 (8GB) cannot run any of the 84 tracked models fully in VRAM at 8k context. It may handle smaller models with CPU offload.
- Can the Apple M3 (8GB) run Llama 3.3 70B Instruct?
- The Apple M3 (8GB) 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 M3 (8GB) run Qwen 3.6 27B?
- The Apple M3 (8GB) does not have enough VRAM to run Qwen 3.6 27B. You would need more VRAM or a lower quantization level.
- Can the Apple M3 (8GB) run Llama 3.1 8B Instruct?
- The Apple M3 (8GB) does not have enough VRAM to run Llama 3.1 8B Instruct. You would need more VRAM or a lower quantization level.