Can I run Marin Mochi 535B A23B locally?
Marin Mochi 535B A23B by open-athena has 535B parameters (23B active per token, 384 experts, 8 used per token). At 4-bit with an 8K context it needs about 308 GB of GPU memory. It fits on 2 of 65 GPUs and Macs we track.
Memory needed by quant and context length
GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use parameter count × bits per weight. How we calculate.
| Quant | Quality | 4K ctx | 8K ctx | 32K ctx | 128K ctx |
|---|---|---|---|---|---|
| 2-bit (Q2_K) | rough | 189 GB | 190 GB | 197 GB | 224 GB |
| 3-bit (Q3_K_M) | usable | 246 GB | 247 GB | 253 GB | 280 GB |
| 4-bit (Q4_K_M) | good, most popular | 307 GB | 308 GB | 314 GB | 341 GB |
| 5-bit (Q5_K_M) | very good | 357 GB | 358 GB | 365 GB | 392 GB |
| 6-bit (Q6_K) | near lossless | 411 GB | 412 GB | 419 GB | 446 GB |
| 8-bit (Q8_0) | effectively lossless | 532 GB | 533 GB | 540 GB | 567 GB |
| 16-bit (F16/BF16) | original weights | 999 GB | 1000 GB | 1007 GB | 1034 GB |
Which GPUs and Macs run Marin Mochi 535B A23B?
Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).
Questions
How much VRAM does Marin Mochi 535B A23B need?
About 308 GB at 4-bit (Q4_K_M) with an 8K context: 305 GB of weights (estimated), 2.3 GB of KV cache and 0.6 GB of runtime overhead. At 8-bit it needs about 533 GB.
What is the smallest GPU that runs Marin Mochi 535B A23B?
No single consumer GPU in our table fits it at Q4_K_M with 8K context; the smallest option that does is the Mac M3 Ultra 512GB (512 GB).
Can I run Marin Mochi 535B A23B on a Mac?
Yes. At Q4_K_M with 8K context it fits on a Mac M3 Ultra 512GB or larger, using the default macOS GPU memory limit.
Similar models
Same family or a similar parameter count, with the VRAM each needs at Q4_K_M and 8K context.
LongCat Flash Chat
562B · 321 GB at 4-bitLing 3.1 Flash 1019
572B · 335 GB at 4-bitLing 3.1 Flash
572B · 335 GB at 4-bitInkling 2 Small
594B · 346 GB at 4-bitMiniMax M1 40K
456B · 263 GB at 4-bitMiniMax Text 01
456B · 263 GB at 4-bitEmbed this
Free badge and "Will it run on my GPU?" widget for model cards, READMEs and blogs. No cookies, no tracking. All options.
Copy a snippet
Markdown badge
[](https://runsonmy.com/model/marin-mochi-535b-a23b/) Widget (script)
<div class="runsonmy-widget" data-model="marin-mochi-535b-a23b"><a href="https://runsonmy.com/model/marin-mochi-535b-a23b/">Will Marin Mochi 535B A23B run on my GPU? Check on RunsOnMy</a></div>
<script async src="https://runsonmy.com/embed.js"></script> Widget (iframe)
<iframe src="https://runsonmy.com/embed/m/marin-mochi-535b-a23b/" width="100%" height="220" style="max-width:440px;border:0" loading="lazy" title="Will Marin Mochi 535B A23B run on my GPU? (RunsOnMy)"></iframe> Model facts
| Hugging Face repo | open-athena/Marin-Mochi-535B-A23B |
| Parameters | 535,477,106,688 (535B) · ~23B active (from the model name) |
| Layers / hidden size | 48 / 6144 |
| Attention | GQA; 48 heads, 12 KV heads, head dim 128 |
| KV cache per token (FP16) | 288.0 KiB |
| Max context | 262,144 tokens |
| License | openmdw-1.1 |
| Released on HF | Nov 3, 2026 |
| Downloads (30d) / likes | 121 / 13 |
Sources: Hugging Face model API · config.json (open-athena/Marin-Mochi-535B-A23B) · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.