What LLMs can a CPU-only PC with 256 GB of quad-channel DDR4-3200 run?
A CPU-only PC with 256 GB of quad-channel DDR4-3200 gives a model 250 GB usable of memory (system RAM minus about 6 GB for the OS and apps) and 102.4 GB/s of bandwidth. Of the 61 open models tracked here, 50 fit at Q4_K_M with an 8,192-token context, and 1 more at a lower precision, each leaving at least 0.5 GB free on this PC.
With no GPU, llama.cpp (or Ollama and LM Studio on top of it) keeps the whole model in the 256 GB of DDR4-3200 and reads all of its active weights over the memory bus for every token. A model fits here when weights, cache and overhead stay within the RAM minus 6 GB held back for the operating system and the apps next to it; close the browser and that reserve shrinks. MoE offload does not apply, since everything already runs from RAM, but mixture-of-experts models still help most: only their active experts are read per token. The speeds use 50–80% of the rated 102.4 GB/s for dense models and 30–55% for MoE models, from public CPU-only llama.cpp runs. Reading the prompt is compute-bound on a CPU, often tens of tokens per second, so a long prompt can take minutes before the first word; a GPU of any size speeds that part up even when the weights stay in RAM.
Models that fit in 256 GB of DDR4-3200 (quad channel, CPU only)
The most precise weights that still fit with 8K tokens of context and one request, the memory that takes, the longest context at that precision and the writing speed. Bigger models come first. gpt-oss-120b (67.7 GB, 9.5–18 tokens/s on this PC) and gpt-oss-20b (14.8 GB, 12–22 tokens/s on this PC) are listed at MXFP4 only, as published.
| Model | Best precision | Memory | Longest context | Tokens/s |
|---|---|---|---|---|
| MiniMax M3 427.0B, 23B active | GGUF Q3_K_M | 215 GB | 272K | 2.5–4.6 |
| GLM-5.3 Flash 321.3B, 18B active | GGUF Q5_K_M | 234 GB | 1M (full) | 2.4–4.4 |
| IQuest-Q1 320B-A15B 320.3B, 15B active | GGUF Q5_K_M | 235 GB | 141K | 2.4–4.4 |
| MiMo V2.6 Flash 310.8B, 15B active | GGUF Q5_K_M | 226 GB | 960K | 2.8–5.1 |
| DeepSeek V4 Flash 0731 304.2B, 13B active | GGUF Q5_K_M | 222 GB | 304K | 3.1–5.7 |
| DeepSeek V4 Flash 290.9B, 13B active | GGUF Q6_K | 246 GB | 49K | 2.7–4.9 |
| MiniMax M2.7 228.7B, 10B active | As published (FP8) | 238 GB | 49K | 2.5–4.6 |
| Step 3.7 Flash 196B-A11B 201.4B, 11B active | GGUF Q8_0 | 220 GB | 256K (full) | 2.5–4.6 |
| Qwen3.8 Flash Next 180.0B, 6B active | GGUF Q8_0 | 197 GB | 256K (full) | 4.6–8.5 |
| Mistral Medium 3.5 128B 127.7B | GGUF Q8_0 | 143 GB | 256K (full) | 0.4–0.6 |
| Qwen3.5 122B-A10B 125.1B, 10B active | GGUF Q8_0 | 137 GB | 256K (full) | 2.8–5.2 |
| Nemotron 3 Super 120B-A12B 123.6B, 12B active | GGUF Q8_0 | 135 GB | 256K (full) | 2.4–4.4 |
| gpt-oss-120b 116.8B, 5.1B active | As published (MXFP4) | 67.7 GB | 128K (full) | 9.5–18 |
| AliceAI Foundation 80B-A3B 81.3B, 3B active | FP16 / BF16 | 167 GB | 256K (full) | 4.9–9.0 |
| Qwen3-Coder-Next 79.7B, 3B active | FP16 / BF16 | 164 GB | 256K (full) | 4.9–9.0 |
| Llama 3.1 70B 70.6B | FP16 / BF16 | 148 GB | 128K (full) | 0.4–0.6 |
| K2-Horizon MoVA 36B-A4B 37.4B, 4B active | FP16 / BF16 | 78.9 GB | 512K (full) | 3.2–5.8 |
| Ornith 1.5 35B-A3B 36.0B, 3B active | FP16 / BF16 | 74.3 GB | 256K (full) | 4.9–9.1 |
| Qwen3.6 35B-A3B 36.0B, 3B active | FP16 / BF16 | 74.3 GB | 256K (full) | 4.9–9.1 |
| Ornith 1.0 35B 35.1B, 3B active | FP16 / BF16 | 72.6 GB | 256K (full) | 4.9–9.1 |
| LLM-jp-4.1 32B-A3B Thinking 32.1B, 3.8B active | FP16 / BF16 | 66.9 GB | 64K (full) | 3.7–6.9 |
| Nemotron 3 Nano 30B-A3B 31.6B, 3.5B active | FP16 / BF16 | 65.3 GB | 256K (full) | 4.3–8.0 |
| Gemma 4 31B 31.3B | FP16 / BF16 | 66.6 GB | 256K (full) | 0.8–1.3 |
| GLM-4.7 Flash 31.2B, 3B active | FP16 / BF16 | 64.9 GB | 198K (full) | 4.7–8.7 |
| Xing 4.0 29B-A4B 31.2B, 4B active | FP16 / BF16 | 64.8 GB | 256K (full) | 3.6–6.7 |
| Qwen3-Coder 30B-A3B 30.5B, 3.3B active | FP16 / BF16 | 63.9 GB | 256K (full) | 4.1–7.6 |
| Qwen3 30B-A3B 30.5B, 3.3B active | FP16 / BF16 | 63.9 GB | 40K (full) | 4.1–7.6 |
| Muse Glimmer 30B 29.8B | FP16 / BF16 | 61.7 GB | 128K (full) | 0.9–1.4 |
| Granite 4.2 30B 29.3B | FP16 / BF16 | 62.7 GB | 128K (full) | 0.8–1.3 |
| Qwen3.6 27B 27.8B | FP16 / BF16 | 58.0 GB | 256K (full) | 0.9–1.5 |
| Qwen3.8 27B 27.8B | FP16 / BF16 | 58.0 GB | 256K (full) | 0.9–1.5 |
| Hemmingway-1 27B 27.3B | FP16 / BF16 | 57.0 GB | 256K (full) | 0.9–1.5 |
| Gemma 4 26B-A4B 25.8B, 4B active | FP16 / BF16 | 53.9 GB | 256K (full) | 3.6–6.6 |
| gpt-oss-20b 20.9B, 3.6B active | As published (MXFP4) | 14.8 GB | 128K (full) | 12–22 |
| Gemma 4 12B 12.0B | FP16 / BF16 | 25.7 GB | 256K (full) | 2.1–3.3 |
| ZDTaichu 5.0 9B 9.8B | FP16 / BF16 | 20.8 GB | 128K (full) | 2.6–4.1 |
| Ornith 1.5 9B 9.7B | FP16 / BF16 | 20.6 GB | 256K (full) | 2.6–4.2 |
| Qwen3.5 9B 9.7B | FP16 / BF16 | 20.6 GB | 256K (full) | 2.6–4.2 |
| Ornith 1.0 9B 9.4B | FP16 / BF16 | 20.1 GB | 256K (full) | 2.7–4.3 |
| MiMo V2.6 Distill Qwen 9B 9.4B | FP16 / BF16 | 20.1 GB | 256K (full) | 2.7–4.3 |
| Granite 4.2 8B 8.8B | FP16 / BF16 | 19.9 GB | 128K (full) | 2.7–4.3 |
| LFM2.5 8B-A1B 8.5B, 1.5B active | FP16 / BF16 | 18.0 GB | 128,000 (full) | 9.8–18 |
| Qwen3 8B 8.2B | FP16 / BF16 | 18.5 GB | 40K (full) | 2.9–4.6 |
| Llama 3.1 8B 8.0B | FP16 / BF16 | 18.1 GB | 128K (full) | 3.0–4.8 |
| Gemma 4 E4B 8.0B | FP16 / BF16 | 17.1 GB | 128K (full) | 3.2–5.1 |
| Ling 3.0 Tiny 7.9B-A1.3B 7.9B, 1.3B active | FP16 / BF16 | 16.7 GB | 128K (full) | 11–21 |
| Spark-X2.5 4B 4.1B | FP16 / BF16 | 9.35 GB | 1M (full) | 5.9–9.4 |
| Nemotron 3 Nano 4B 4.0B | FP16 / BF16 | 8.78 GB | 256K (full) | 6.3–10 |
| Granite 4.2 3B 3.7B | FP16 / BF16 | 8.69 GB | 128K (full) | 6.3–10 |
| MiniCPM5 2B 2.5B | FP16 / BF16 | 6.02 GB | 128K (full) | 9.4–15 |
| Limite 1B Violetto 1.0B | FP16 / BF16 | 2.79 GB | 128K (full) | 22–36 |
What one GPU adds to this PC
With a graphics card, llama.cpp's --n-cpu-moe keeps the routed experts of the first layers in this PC's
256 GB of DDR4-3200 and everything else on the card, so a mixture-of-experts model can run far bigger than the card and far faster than
the CPU alone. Each measured GGUF file below at its own 32K context (or its whole window): tokens per second on the CPU alone
at MXFP4 where it fits, then the smallest --n-cpu-moe and tokens per second with an
RTX 5060 Ti 16GB or an RTX 4090, the experts read at 102.4 GB/s.
"More RAM" means the experts left on the CPU need more than the 250 GB this PC leaves for a model.
| Model | GGUF | CPU only | --n-cpu-moe, RTX 5060 Ti 16GB | --n-cpu-moe, RTX 4090 | Tokens/s, RTX 5060 Ti 16GB | Tokens/s, RTX 4090 |
|---|---|---|---|---|---|---|
| gpt-oss-20b | MXFP4 | 9.5–17 | 0 | 0 | 37–64 | 78–138 |
| Gemma 4 26B-A4B | UD-Q4_K_M | 8.9–16 | 3 | 0 | 32–55 | 73–128 |
| Qwen3 30B-A3B | Q4_K_M | 5.8–11 | 15 | 0 | 21–35 | 54–93 |
| GLM-4.7 Flash | Q4_K_M | 8.5–16 | 12 | 0 | 26–44 | 65–114 |
| Nemotron 3 Nano 30B-A3B | Q4_K_M | 13–24 | 21 | 0 | 30–52 | 94–167 |
| Ornith 1.5 35B-A3B | Q4_K_M | 12–22 | 13 | 0 | 38–65 | 96–171 |
| Qwen3.6 35B-A3B | UD-Q4_K_M | 12–22 | 13 | 0 | 32–55 | 81–142 |
| Qwen3-Coder-Next | Q4_K_M | 12–21 | 35 | 26 | 24–40 | 37–64 |
| gpt-oss-120b | MXFP4 | 7.4–14 | 29 | 24 | 13–22 | 18–31 |
| Qwen3.5 122B-A10B | Q4_K_M | 4.5–8.2 | 42 | 36 | 9–15 | 13–22 |
| Qwen3.8 Flash Next | UD-Q4_K_XL | 6.9–13 | 42 | 37 | 11–19 | 17–29 |
Too big for 256 GB of DDR4-3200 (quad channel, CPU only) at 8K context
How much memory each of these lacks at Q4_K_M with 8K context, one request and 0.5 GB kept free, out of the 250 GB this PC leaves for a model; more RAM or a smaller quant is the way in.
- DeepSeek V3 / R1: 176 GB short
- DeepSeek V3.2: 176 GB short
- GLM-5.3: 219 GB short
- GLM-5.2: 219 GB short
- DeepSeek V4.1 Flash: 225 GB short
- Hy4 Preview 770B-A49B: 235 GB short
- MiMo V2.6 Pro: 386 GB short
- DeepSeek V4 Pro: 743 GB short
- Qwen3.8 2.4T-A95B: 1,268 GB short
- Kimi K3: 1,474 GB short
Best models for 256 GB of DDR4-3200 (quad channel, CPU only) with room for long context
The biggest models that still fit at Q4_K_M with 32,768 tokens of context (or their whole window, if shorter), the memory left over, the longest context the PC takes at that precision and the writing speed with that context in the cache, from 102.4 GB/s of bandwidth.
| Model | Context | Memory | Left over | Longest context | Tokens/s |
|---|---|---|---|---|---|
| GLM-5.3 Flash 321.3B | 32K | 200 GB | 49.9 GB | 1M (full) | 2.7–5.0 |
| IQuest-Q1 320B-A15B 320.3B | 32K | 204 GB | 46.3 GB | 458K | 2.2–4.1 |
| MiMo V2.6 Flash 310.8B | 32K | 194 GB | 56.0 GB | 1M (full) | 3.1–5.6 |
| DeepSeek V4 Flash 0731 304.2B | 32K | 192 GB | 58.0 GB | 654K | 2.8–5.2 |
| DeepSeek V4 Flash 290.9B | 32K | 184 GB | 66.2 GB | 743K | 2.8–5.2 |
How fast this PC writes as the context fills
Tokens per second at Q4_K_M for the biggest models that fit with 32K tokens, with 8K and 32K tokens in the cache and at the longest context the PC holds, and at Q8_0 with 8K where that fits. Each token reads the active weights and the whole cache once, so the speed follows the 102.4 GB/s of bandwidth and falls as the cache grows. The ceiling is that bandwidth divided by the bytes read per token, which no runtime reaches; the reply time is for 1,000 new tokens with 32K (or the longest context) in the cache, prompt processing aside.
| Model | 8K | 32K | Longest | At the longest | Q8_0 at 8K | Bandwidth ceiling at 8K | 1,000-token reply at 32K |
|---|---|---|---|---|---|---|---|
| GLM-5.3 Flash | 2.8–5.1 | 2.7–5.0 | 1M | 1.3–2.3 | Does not fit | 9 | 201–370 s |
| IQuest-Q1 320B-A15B | 2.8–5.1 | 2.2–4.1 | 458K | 0.5–1.0 | Does not fit | 9 | 242–445 s |
| MiMo V2.6 Flash | 3.2–6.0 | 3.1–5.6 | 1M | 0.9–1.7 | Does not fit | 11 | 178–326 s |
| DeepSeek V4 Flash 0731 | 3.6–6.5 | 2.8–5.2 | 654K | 0.5–0.8 | Does not fit | 12 | 191–351 s |
| DeepSeek V4 Flash | 3.6–6.5 | 2.8–5.2 | 743K | 0.4–0.8 | Does not fit | 12 | 191–351 s |
| MiniMax M2.7 | 3.8–6.9 | 2.1–3.9 | 200K | 0.5–1.0 | Does not fit | 13 | 256–469 s |
| Step 3.7 Flash 196B-A11B | 4.2–7.8 | 3.6–6.7 | 256K | 1.6–2.9 | 2.5–4.6 | 14 | 150–275 s |
| Qwen3.8 Flash Next | 7.9–15 | 6.9–13 | 256K | 3.0–5.6 | 4.6–8.5 | 27 | 79–146 s |
| Mistral Medium 3.5 128B | 0.6–1.0 | 0.6–0.9 | 256K | 0.3–0.5 | 0.4–0.6 | 1 | 1088–1741 s |
| Qwen3.5 122B-A10B | 4.9–9.0 | 4.5–8.2 | 256K | 2.5–4.5 | 2.8–5.2 | 16 | 122–225 s |
Near misses on 256 GB of DDR4-3200 (quad channel, CPU only)
Models that miss at Q4_K_M with 32,768 tokens of context (or their whole window), or leave under 0.5 GB free there, but fit with a shorter context or 3-bit weights, or are short by at most a quarter of the memory and fit with MoE offload. Smallest shortfall first.
| Model | Short by at Q4_K_M | Fits instead |
|---|---|---|
| MiniMax M3 427.0B | 19.3 GB | GGUF Q3_K_M with 32K, 218 GB |
How these numbers are worked out
Memory is the weights at each precision plus the KV cache for 8,192 tokens in FP16 and runtime overhead (0.5 GB plus 10%), computed from each model's files on Hugging Face with the LLM VRAM Calculator. Speed is estimated from memory bandwidth and the parameters read per token with the LLM Speed Calculator. Both tools take any other model from Hugging Face.
Nearby GPUs and setups
How many of the tracked models each one holds at Q4_K_M with 8K context, against 50 here.
| Setup | How it relates | Memory | Bandwidth | Models at Q4_K_M |
|---|---|---|---|---|
| CPU only, DDR4-3200 quad channel (128 GB) | Next size down | 122 GB usable | 102.4 GB/s | 43 (−7) |
Model numbers read from Hugging Face on .