Apple Silicon (Mac)
Apple M4 Pro
Data as of
- Memory bandwidth
- 273 GB/s
- Source
- Checked:
With 24 GB
One entry per model, up to 30, sorted by popularity. For each model we show the largest (most precise) quantization that fits in this memory with a context of at least 8K tokens. The context length is the longest one that quantization fits (for models whose maximum is under 8K, such as embedding models, that maximum). Models that are not for chat or text generation are labelled with their use. "Left" is the memory minus the estimate; under 5% of the memory is marked "tight" (not a guarantee that it fits).
-
Context length 8,192 Estimated 25.6 GB 0.17 GB left (tight) Speed 6.9–9.2
-
Ornith-1.5-9B BF16Context length 131,072 Estimated 23.8 GB 1.94 GB left Speed 8.9–12
-
Context length 131,072 Estimated 25.5 GB 0.26 GB left (tight) Speed not estimated (MoE)
-
Context length 65,536 Estimated 24.9 GB 0.87 GB left (tight) Speed not estimated (MoE)
-
Context length 8,192 Estimated 25.7 GB 0.09 GB left (tight) Speed 6.8–9.1
-
Context length 262,144 Estimated 23.6 GB 2.22 GB left Speed not estimated (MoE)
-
Context length 65,536 Estimated 22.6 GB 3.15 GB left Speed 9.6–13
-
Context length 65,536 Estimated 25.2 GB 0.57 GB left (tight) Speed not estimated (MoE)
-
Context length 16,384 Estimated 25.7 GB 0.09 GB left (tight) Speed 6.9–9.2
-
Context length 16,384 Estimated 24.7 GB 1.06 GB left (tight) Speed 7.3–9.7
-
Ornith-1.0-9B BF16Context length 131,072 Estimated 23.3 GB 2.43 GB left Speed 9.1–12
-
Context length 512 Estimated 2.41 GB 23.4 GB left Speed 123–163
-
Context length 65,536 Estimated 25.3 GB 0.46 GB left (tight) Speed not estimated (MoE)
-
Context length 131,072 Estimated 23.8 GB 1.97 GB left Speed 8.9–12
-
Context length 40,960 Estimated 23.3 GB 2.49 GB left Speed 10.0–13
-
Context length 65,536 Estimated 25.7 GB 0.05 GB left (tight) Speed 8.1–11
-
Context length 32,768 Estimated 25.4 GB 0.42 GB left (tight) Speed not estimated (MoE)
-
Context length 32,768 Estimated 22.1 GB 3.70 GB left Speed 10.0–13
-
Context length 131,072 Estimated 23.8 GB 1.97 GB left Speed 8.9–12
-
Context length 131,072 Estimated 23.3 GB 2.46 GB left Speed 9.1–12
-
Context length 65,536 Estimated 25.3 GB 0.45 GB left (tight) Speed not estimated (MoE)
-
Ternary-Bonsai-2-27B Q2_0Context length 131,072 Estimated 17.4 GB 8.35 GB left Speed 21–29
-
Context length 131,072 Estimated 21.9 GB 3.86 GB left Speed 14–18
-
Context length 262,144 Estimated 18.4 GB 7.42 GB left Speed 19–25
-
Ornith-1.0-35B Q4_K_MContext length 131,072 Estimated 25.0 GB 0.78 GB left (tight) Speed not estimated (MoE)
-
Context length 8,192 Estimated 25.3 GB 0.45 GB left (tight) Speed 6.9–9.3
-
Context length 32,768 Estimated 25.3 GB 0.44 GB left (tight) Speed 7.5–9.9
-
Wan2.1 Q4_K_MContext length 65,536 Estimated 22.3 GB 3.43 GB left Speed 9.7–13
-
Context length 131,072 Estimated 21.9 GB 3.86 GB left Speed 14–18
-
Context length 65,536 Estimated 12.5 GB 13.3 GB left Speed 40–53
With 48 GB
One entry per model, up to 30, sorted by popularity. For each model we show the largest (most precise) quantization that fits in this memory with a context of at least 8K tokens. The context length is the longest one that quantization fits (for models whose maximum is under 8K, such as embedding models, that maximum). Models that are not for chat or text generation are labelled with their use. "Left" is the memory minus the estimate; under 5% of the memory is marked "tight" (not a guarantee that it fits).
-
Qwen3.8-27B UD-Q8_K_XLContext length 262,144 Estimated 49.8 GB 1.70 GB left (tight) Speed 5.2–6.9
-
Ornith-1.5-9B BF16Context length 262,144 Estimated 28.1 GB 23.4 GB left Speed 8.9–12
-
Context length 262,144 Estimated 44.3 GB 7.26 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 44.3 GB 7.25 GB left Speed not estimated (MoE)
-
Context length 131,072 Estimated 48.6 GB 2.97 GB left Speed 4.2–5.6
-
Context length 262,144 Estimated 33.6 GB 17.9 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 35.5 GB 16.0 GB left Speed 9.6–13
-
Context length 262,144 Estimated 45.3 GB 6.22 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 29.7 GB 21.8 GB left Speed 6.9–9.2
-
Context length 262,144 Estimated 47.4 GB 4.11 GB left Speed 5.6–7.5
-
Ornith-1.0-9B BF16Context length 262,144 Estimated 27.6 GB 23.9 GB left Speed 9.1–12
-
Context length 512 Estimated 2.41 GB 49.1 GB left Speed 123–163
-
Context length 262,144 Estimated 45.6 GB 5.96 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 23.4 GB left Speed 8.9–12
-
Context length 40,960 Estimated 23.3 GB 28.3 GB left Speed 10.0–13
-
Context length 262,144 Estimated 49.8 GB 1.70 GB left (tight) Speed 5.2–6.9
-
Context length 262,144 Estimated 34.4 GB 17.1 GB left Speed not estimated (MoE)
-
Context length 131,072 Estimated 36.6 GB 15.0 GB left Speed 10.0–13
-
Context length 262,144 Estimated 28.1 GB 23.4 GB left Speed 8.9–12
-
Context length 262,144 Estimated 27.6 GB 23.9 GB left Speed 9.1–12
-
Context length 262,144 Estimated 44.9 GB 6.61 GB left Speed not estimated (MoE)
-
Ternary-Bonsai-2-27B Q2_0Context length 262,144 Estimated 26.0 GB 25.5 GB left Speed 21–29
-
Context length 262,144 Estimated 30.5 GB 21.0 GB left Speed 14–18
-
Context length 262,144 Estimated 18.4 GB 33.2 GB left Speed 19–25
-
Ornith-1.0-35B Q8_0Context length 262,144 Estimated 43.4 GB 8.13 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 48.6 GB 2.92 GB left Speed 5.4–7.2
-
Context length 262,144 Estimated 49.8 GB 1.70 GB left (tight) Speed 5.2–6.9
-
Wan2.1 IQ4_NLContext length 262,144 Estimated 47.2 GB 4.33 GB left Speed 5.7–7.6
-
Context length 262,144 Estimated 30.5 GB 21.0 GB left Speed 14–18
-
Context length 65,536 Estimated 12.5 GB 39.1 GB left Speed 40–53
Frequently asked questions
- Which local LLMs run on Apple M4 Pro?
- On this site, 981 models have a quantization that fits in 48 GB at 8,192 tokens (estimated). Popular ones include Qwen3.8-27B, Ornith-1.5-9B, Huihui-Qwen3.6-35B-A3B-abliterated-MTP, Ornith-1.5-35B-A3B, Huihui-Qwen3.8-27B-abliterated.
- What is the largest model that runs on Apple M4 Pro?
- Among the models listed here, GLM-5.3-Flash-Coder-RCO-152E (about 169.6B); IQ2_XXS fits in 48 GB (estimated at 8,192 tokens).
- How fast does Apple M4 Pro generate text?
- With 273 GB/s of memory bandwidth, Qwen3.8-27B at UD-Q8_K_XL (31.5 GB) runs at an estimated 5.2–6.9 tokens per second (60–80% of bandwidth ÷ file size; prompt processing not included).