Apple Silicon (Mac)
Apple M2 Ultra
Data as of
- Memory bandwidth
- 800 GB/s
- Source
- Checked:
With 64 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 32,768 Estimated 68.5 GB 0.23 GB left (tight) Speed not estimated (MoE)
-
Qwen3.8-27B BF16Context length 131,072 Estimated 64.4 GB 4.27 GB left Speed 8.8–12
-
Ornith-1.5-9B BF16Context length 262,144 Estimated 28.1 GB 40.6 GB left Speed 26–35
-
Context length 262,144 Estimated 44.3 GB 24.4 GB left Speed not estimated (MoE)
-
Context length 16,384 Estimated 68.4 GB 0.28 GB left (tight) Speed not estimated (MoE)
-
Context length 262,144 Estimated 44.3 GB 24.4 GB left Speed not estimated (MoE)
-
Context length 131,072 Estimated 64.4 GB 4.27 GB left Speed 8.8–12
-
Context length 262,144 Estimated 33.6 GB 35.1 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 35.5 GB 33.2 GB left Speed 28–37
-
Context length 262,144 Estimated 45.3 GB 23.4 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 29.7 GB 39.0 GB left Speed 20–27
-
Context length 262,144 Estimated 47.4 GB 21.3 GB left Speed 17–22
-
Ornith-1.0-9B BF16Context length 262,144 Estimated 27.6 GB 41.1 GB left Speed 27–36
-
Context length 512 Estimated 2.41 GB 66.3 GB left Speed 359–479
-
Context length 262,144 Estimated 45.6 GB 23.1 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 40.6 GB left Speed 26–35
-
Context length 40,960 Estimated 23.3 GB 45.4 GB left Speed 29–39
-
Context length 262,144 Estimated 49.8 GB 18.9 GB left Speed 15–20
-
Context length 262,144 Estimated 57.3 GB 11.5 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 55.9 GB 12.8 GB left Speed 29–39
-
Context length 32,768 Estimated 68.4 GB 0.27 GB left (tight) Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 40.6 GB left Speed 26–35
-
Context length 262,144 Estimated 27.6 GB 41.1 GB left Speed 27–36
-
Context length 262,144 Estimated 44.9 GB 23.8 GB left Speed not estimated (MoE)
-
Context length 131,072 Estimated 63.6 GB 5.12 GB left Speed 8.9–12
-
Context length 262,144 Estimated 30.5 GB 38.2 GB left Speed 40–53
-
Context length 262,144 Estimated 18.4 GB 50.4 GB left Speed 55–74
-
Ornith-1.0-35B Q8_0Context length 262,144 Estimated 43.4 GB 25.3 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 48.6 GB 20.1 GB left Speed 16–21
-
Context length 262,144 Estimated 49.8 GB 18.9 GB left Speed 15–20
With 128 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 262,144 Estimated 92.0 GB 45.4 GB left Speed not estimated (MoE)
-
Qwen3.8-27B BF16Context length 262,144 Estimated 73.0 GB 64.4 GB left Speed 8.8–12
-
Ornith-1.5-9B BF16Context length 262,144 Estimated 28.1 GB 109 GB left Speed 26–35
-
Context length 262,144 Estimated 77.5 GB 59.9 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 92.3 GB 45.2 GB left Speed not estimated (MoE)
-
Ornith-1.5-35B-A3B BF16Context length 262,144 Estimated 77.6 GB 59.9 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 73.0 GB 64.4 GB left Speed 8.8–12
-
Context length 262,144 Estimated 33.6 GB 104 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 35.5 GB 102 GB left Speed 28–37
-
Context length 262,144 Estimated 45.3 GB 92.1 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 29.7 GB 108 GB left Speed 20–27
-
Context length 262,144 Estimated 47.4 GB 90.0 GB left Speed 17–22
-
Ornith-1.0-9B BF16Context length 262,144 Estimated 27.6 GB 110 GB left Speed 27–36
-
Context length 512 Estimated 2.41 GB 135 GB left Speed 359–479
-
Context length 262,144 Estimated 77.5 GB 59.9 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 109 GB left Speed 26–35
-
Context length 40,960 Estimated 23.3 GB 114 GB left Speed 29–39
-
Context length 262,144 Estimated 49.8 GB 87.6 GB left Speed 15–20
-
Context length 262,144 Estimated 57.3 GB 80.2 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 55.9 GB 81.5 GB left Speed 29–39
-
Context length 32,768 Estimated 137 GB 0.70 GB left (tight) Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 109 GB left Speed 26–35
-
Context length 262,144 Estimated 27.6 GB 110 GB left Speed 27–36
-
Context length 262,144 Estimated 77.5 GB 59.9 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 72.2 GB 65.2 GB left Speed 8.9–12
-
Context length 262,144 Estimated 30.5 GB 107 GB left Speed 40–53
-
Context length 262,144 Estimated 18.4 GB 119 GB left Speed 55–74
-
Ornith-1.0-35B BF16Context length 262,144 Estimated 75.9 GB 61.6 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 48.6 GB 88.8 GB left Speed 16–21
-
Context length 262,144 Estimated 49.8 GB 87.6 GB left Speed 15–20
With 192 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 262,144 Estimated 92.0 GB 114 GB left Speed not estimated (MoE)
-
Qwen3.8-27B BF16Context length 262,144 Estimated 73.0 GB 133 GB left Speed 8.8–12
-
Ornith-1.5-9B BF16Context length 262,144 Estimated 28.1 GB 178 GB left Speed 26–35
-
Context length 262,144 Estimated 77.5 GB 129 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 92.3 GB 114 GB left Speed not estimated (MoE)
-
Ornith-1.5-35B-A3B BF16Context length 262,144 Estimated 77.6 GB 129 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 73.0 GB 133 GB left Speed 8.8–12
-
Context length 262,144 Estimated 33.6 GB 173 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 35.5 GB 171 GB left Speed 28–37
-
Context length 262,144 Estimated 45.3 GB 161 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 29.7 GB 176 GB left Speed 20–27
-
Context length 262,144 Estimated 47.4 GB 159 GB left Speed 17–22
-
Ornith-1.0-9B BF16Context length 262,144 Estimated 27.6 GB 179 GB left Speed 27–36
-
Context length 512 Estimated 2.41 GB 204 GB left Speed 359–479
-
Context length 262,144 Estimated 77.5 GB 129 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 178 GB left Speed 26–35
-
Context length 40,960 Estimated 23.3 GB 183 GB left Speed 29–39
-
Context length 262,144 Estimated 49.8 GB 156 GB left Speed 15–20
-
Context length 262,144 Estimated 57.3 GB 149 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 55.9 GB 150 GB left Speed 29–39
-
Context length 262,144 Estimated 197 GB 9.49 GB left (tight) Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 178 GB left Speed 26–35
-
Context length 262,144 Estimated 27.6 GB 179 GB left Speed 27–36
-
Context length 262,144 Estimated 77.5 GB 129 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 72.2 GB 134 GB left Speed 8.9–12
-
Context length 262,144 Estimated 30.5 GB 176 GB left Speed 40–53
-
Context length 262,144 Estimated 18.4 GB 188 GB left Speed 55–74
-
Ornith-1.0-35B BF16Context length 262,144 Estimated 75.9 GB 130 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 48.6 GB 158 GB left Speed 16–21
-
Context length 262,144 Estimated 49.8 GB 156 GB left Speed 15–20
Frequently asked questions
- Which local LLMs run on Apple M2 Ultra?
- On this site, 1,043 models have a quantization that fits in 192 GB at 8,192 tokens (estimated). Popular ones include Qwen3.8-27B, Qwen3.8-Flash-Next-GSQ-RCO, Ornith-1.5-9B, Huihui-Qwen3.6-35B-A3B-abliterated-MTP, Ornith-1.5-35B-A3B.
- What is the largest model that runs on Apple M2 Ultra?
- Among the models listed here, DeepSeek-V3-0324 (about 671B); IQ1_M fits in 192 GB (estimated at 8,192 tokens).
- How fast does Apple M2 Ultra generate text?
- With 800 GB/s of memory bandwidth, Qwen3.8-27B at BF16 (54.7 GB) runs at an estimated 8.8–12 tokens per second (60–80% of bandwidth ÷ file size; prompt processing not included).