Apple Silicon (Mac)
Apple M5 Ultra
Data as of
- Memory bandwidth
- 1,200 GB/s
- Source
- Checked:
With 96 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 11.1 GB left Speed not estimated (MoE)
-
Qwen3.8-27B BF16Context length 262,144 Estimated 73.0 GB 30.0 GB left Speed 13–18
-
Ornith-1.5-9B BF16Context length 262,144 Estimated 28.1 GB 75.0 GB left Speed 39–52
-
Context length 262,144 Estimated 77.5 GB 25.5 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 92.3 GB 10.8 GB left Speed not estimated (MoE)
-
Ornith-1.5-35B-A3B BF16Context length 262,144 Estimated 77.6 GB 25.5 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 73.0 GB 30.0 GB left Speed 13–18
-
Context length 262,144 Estimated 33.6 GB 69.5 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 35.5 GB 67.6 GB left Speed 42–56
-
Context length 262,144 Estimated 45.3 GB 57.8 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 29.7 GB 73.4 GB left Speed 30–40
-
Context length 262,144 Estimated 47.4 GB 55.6 GB left Speed 25–33
-
Ornith-1.0-9B BF16Context length 262,144 Estimated 27.6 GB 75.4 GB left Speed 40–54
-
Context length 512 Estimated 2.41 GB 101 GB left Speed 538–718
-
Context length 262,144 Estimated 77.5 GB 25.5 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 75.0 GB left Speed 39–52
-
Context length 40,960 Estimated 23.3 GB 79.8 GB left Speed 44–59
-
Context length 262,144 Estimated 49.8 GB 53.2 GB left Speed 23–31
-
Context length 262,144 Estimated 57.3 GB 45.8 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 55.9 GB 47.2 GB left Speed 44–59
-
Context length 32,768 Estimated 103 GB 0.41 GB left (tight) Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 75.0 GB left Speed 39–52
-
Context length 262,144 Estimated 27.6 GB 75.5 GB left Speed 40–54
-
Context length 262,144 Estimated 77.5 GB 25.5 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 72.2 GB 30.9 GB left Speed 13–18
-
Context length 262,144 Estimated 30.5 GB 72.6 GB left Speed 59–79
-
Context length 262,144 Estimated 18.4 GB 84.7 GB left Speed 83–111
-
Ornith-1.0-35B BF16Context length 262,144 Estimated 75.9 GB 27.2 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 48.6 GB 54.5 GB left Speed 24–32
-
Context length 262,144 Estimated 49.8 GB 53.2 GB left Speed 23–31
With 256 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 183 GB left Speed not estimated (MoE)
-
Qwen3.8-27B BF16Context length 262,144 Estimated 73.0 GB 202 GB left Speed 13–18
-
Ornith-1.5-9B BF16Context length 262,144 Estimated 28.1 GB 247 GB left Speed 39–52
-
Context length 262,144 Estimated 77.5 GB 197 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 92.3 GB 183 GB left Speed not estimated (MoE)
-
Ornith-1.5-35B-A3B BF16Context length 262,144 Estimated 77.6 GB 197 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 73.0 GB 202 GB left Speed 13–18
-
Context length 262,144 Estimated 33.6 GB 241 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 35.5 GB 239 GB left Speed 42–56
-
Context length 262,144 Estimated 45.3 GB 230 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 29.7 GB 245 GB left Speed 30–40
-
Context length 262,144 Estimated 47.4 GB 227 GB left Speed 25–33
-
Ornith-1.0-9B BF16Context length 262,144 Estimated 27.6 GB 247 GB left Speed 40–54
-
Context length 512 Estimated 2.41 GB 272 GB left Speed 538–718
-
Context length 262,144 Estimated 77.5 GB 197 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 247 GB left Speed 39–52
-
Context length 40,960 Estimated 23.3 GB 252 GB left Speed 44–59
-
Context length 262,144 Estimated 49.8 GB 225 GB left Speed 23–31
-
Context length 262,144 Estimated 57.3 GB 218 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 55.9 GB 219 GB left Speed 44–59
-
Context length 262,144 Estimated 197 GB 78.2 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 247 GB left Speed 39–52
-
Context length 262,144 Estimated 27.6 GB 247 GB left Speed 40–54
-
Context length 262,144 Estimated 77.5 GB 197 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 72.2 GB 203 GB left Speed 13–18
-
Context length 262,144 Estimated 30.5 GB 244 GB left Speed 59–79
-
Context length 262,144 Estimated 18.4 GB 257 GB left Speed 83–111
-
Ornith-1.0-35B BF16Context length 262,144 Estimated 75.9 GB 199 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 48.6 GB 226 GB left Speed 24–32
-
Context length 262,144 Estimated 49.8 GB 225 GB left Speed 23–31
With 512 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 458 GB left Speed not estimated (MoE)
-
Qwen3.8-27B BF16Context length 262,144 Estimated 73.0 GB 477 GB left Speed 13–18
-
Ornith-1.5-9B BF16Context length 262,144 Estimated 28.1 GB 522 GB left Speed 39–52
-
Context length 262,144 Estimated 77.5 GB 472 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 92.3 GB 457 GB left Speed not estimated (MoE)
-
Ornith-1.5-35B-A3B BF16Context length 262,144 Estimated 77.6 GB 472 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 73.0 GB 477 GB left Speed 13–18
-
Context length 262,144 Estimated 33.6 GB 516 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 35.5 GB 514 GB left Speed 42–56
-
Context length 262,144 Estimated 45.3 GB 504 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 29.7 GB 520 GB left Speed 30–40
-
Context length 262,144 Estimated 47.4 GB 502 GB left Speed 25–33
-
Ornith-1.0-9B BF16Context length 262,144 Estimated 27.6 GB 522 GB left Speed 40–54
-
Context length 512 Estimated 2.41 GB 547 GB left Speed 538–718
-
Context length 262,144 Estimated 77.5 GB 472 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 522 GB left Speed 39–52
-
Context length 40,960 Estimated 23.3 GB 526 GB left Speed 44–59
-
Context length 262,144 Estimated 49.8 GB 500 GB left Speed 23–31
-
Context length 262,144 Estimated 57.3 GB 492 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 55.9 GB 494 GB left Speed 44–59
-
Context length 262,144 Estimated 362 GB 187 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 28.1 GB 522 GB left Speed 39–52
-
Context length 262,144 Estimated 27.6 GB 522 GB left Speed 40–54
-
Context length 262,144 Estimated 77.5 GB 472 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 72.2 GB 478 GB left Speed 13–18
-
Context length 262,144 Estimated 30.5 GB 519 GB left Speed 59–79
-
Context length 262,144 Estimated 18.4 GB 531 GB left Speed 83–111
-
Ornith-1.0-35B BF16Context length 262,144 Estimated 75.9 GB 474 GB left Speed not estimated (MoE)
-
Context length 262,144 Estimated 48.6 GB 501 GB left Speed 24–32
-
Context length 262,144 Estimated 49.8 GB 500 GB left Speed 23–31
Frequently asked questions
- Which local LLMs run on Apple M5 Ultra?
- On this site, 1,052 models have a quantization that fits in 512 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 M5 Ultra?
- Among the models listed here, Kimi-K3 (about 2,779.5B); UD-TQ1_0 fits in 512 GB (estimated at 8,192 tokens).
- How fast does Apple M5 Ultra generate text?
- With 1,200 GB/s of memory bandwidth, Qwen3.8-27B at BF16 (54.7 GB) runs at an estimated 13–18 tokens per second (60–80% of bandwidth ÷ file size; prompt processing not included).