GGPUfit

Hardware

CPU only · DDR4-3200

CPU only · DDR4-3200 has 8 / 16 / 32 / 64 / 128 GB System RAM and 51.2 GB/s. See which open models fit, at which quant, and roughly how fast. Figures below use the default memory (32 GB). Updated 2026-10-17.

Memory8 / 16 / 32 / 64 / 128 GB System RAM
Bandwidth51.2 GB/s
Kindcpu
Year2024

Dual-channel DDR4-3200 theoretical bandwidth: 2 × 3.2 GT/s × 8 bytes = 51.2 GB/s. Source.

RSS for new models on this device

Models that fit

ModelVerdictQuantContexttok/s
Qwen3.8 35B A3B Distill
35.5B
Runs on CPUQ5_K_M8K12–17
Qwen3.6 35B A3B
34.7B
Runs on CPUUD-Q5_K_S8K12–17
Qwen3.5 35B A3B
34.7B
Runs on CPUQ5_K_S8K12–17
Nemotron 3 Nano 30B A3B
31.6B
Runs on CPUQ5_K_S8K10–14
NVIDIA Nemotron 3 Nano Omni 30B A3B Reasoning
31.6B
Runs on CPUUD-Q5_K_S8K10–14
NVIDIA Nemotron 3.5 Lightning 30B A3B
32.9B
Runs on CPUUD-Q4_K_XL4K11–15
LFM2.5 24B A2B
24.0B
Runs on CPUQ8_04K11–16
Qwen3 Next 80B A3B Instruct
79.7B
Runs on CPUUD-IQ1_M8K16–22
Qwen3 Next 80B A3B Thinking
79.7B
Runs on CPUUD-IQ1_M8K16–22
LFM2.5 8B A1B
8.5B
Runs on CPUBF168K12–16
Llama 3.2 1B Instruct
1.2B
Runs on CPUBF168K10–14
LFM2.5 1.2B Instruct
1.2B
Runs on CPUBF168K11–15
gemma 3 1b it
1.0B
Runs on CPUBF168K13–17
Qwen3.5 0.8B
0.75B
Runs on CPUBF168K17–24
Qwen3 0.6B
0.60B
Runs on CPUBF168K13–18
Qwen2.5 0.5B Instruct
0.49B
Runs on CPUF168K26–35
Qwen2.5 32B Instruct
32.8B
Runs on CPUQ5_K_M8K1.1–1.5
Qwen3 32B
32.8B
Runs on CPUUD-Q5_K_XL8K1.1–1.5
Qwen2.5 Coder 32B Instruct
32.8B
Runs on CPUQ5_K_M8K1.1–1.5
DeepSeek R1 Distill Qwen 32B
32.8B
Runs on CPUQ5_K_M8K1.1–1.5
Qwen3 VL 32B Instruct
32.8B
Runs on CPUUD-Q5_K_XL8K1.1–1.5
GLM 4 32B 0414
32.6B
Runs on CPUUD-Q5_K_XL8K1.2–1.6
gemma 4 31B it
30.7B
Runs on CPUUD-Q5_K_XL8K0.9–1.3
Qwen3.8 27B
27.3B
Runs on CPUUD-Q6_K_XL8K1.1–1.5
gemma 3 27b it
27.0B
Runs on CPUUD-Q6_K_XL8K1.0–1.4
medgemma 27b text it
27.0B
Runs on CPUUD-Q6_K_XL8K1.0–1.4
Qwen3.6 27B
26.9B
Runs on CPUQ6_K8K1.2–1.7
Qwen3.5 27B
26.9B
Runs on CPUQ6_K8K1.2–1.7
Kwaipilot KAT Coder V2.5 Dev
34.7B
Runs on CPUQ5_K_M8K6.0–8.2
Devstral Small 2 24B Instruct 2512
23.6B
Runs on CPUQ8_04K1.1–1.5
Mistral Small 3.2 24B Instruct 2506
23.6B
Runs on CPUQ8_04K1.1–1.5
Magistral Small 2509
23.6B
Runs on CPUQ8_04K1.1–1.5
GLM 4.7 Flash
29.9B
Runs on CPUQ6_K8K4.9–6.7
Qwen3 Coder 30B A3B Instruct
30.5B
Runs on CPUQ6_K8K8.6–12
Qwen3 30B A3B Instruct 2507
30.5B
Runs on CPUQ6_K8K8.6–12
Qwen3 VL 30B A3B Instruct
30.5B
Runs on CPUQ6_K8K8.6–12
Qwen3 30B A3B Thinking 2507
30.5B
Runs on CPUQ6_K8K8.6–12
Qwen2.5 72B Instruct
72.7B
Runs on CPUIQ1_M4K1.1–1.5
Llama 3.3 70B Instruct
70.6B
Runs on CPUUD-IQ2_XXS8K1.3–1.7
DeepSeek R1 Distill Llama 70B
70.6B
Runs on CPUUD-IQ2_XXS8K1.3–1.7

90 more models in the catalog do not fit this default profile, or ranked below the first 40 that do.

Example: Runs on CPU on your CPU only · DDR4-3200 at Q5_K_M, 8K context, about 12–17 tok/s. Weights 23.6 GB plus 0.19 GB of KV cache — 24.3 GB of 25.2 GB usable. Up to 16K context still fits. Weights shown in the verdict use GiB (the same 1024³ counting as GPU memory).