Qwen3.8-27B-AP requirements — can you run it?
~27.0B parameters · 8 quantizations measured · updated 2026-09-30 · source: agentionai/Qwen3.8-27B-AP-GGUF
# llama.cpp (auto-downloads the GGUF) llama-cli -hf agentionai/Qwen3.8-27B-AP-GGUF:IQ2_S # Ollama (pulls straight from Hugging Face) ollama run hf.co/agentionai/Qwen3.8-27B-AP-GGUF:IQ2_S # LM Studio: search "agentionai/Qwen3.8-27B-AP-GGUF" in the model browser
Commands load the exact quant measured on this page.
Every quantization: real file size vs what you actually need
| Quant | File (measured) | VRAM/RAM needed (8K ctx) | Runs on (first fit) |
|---|---|---|---|
| IQ2_S | 9.6 GB | 13.0 GB | RTX 4060 Ti 16GB |
| IQ3_XXS | 10.7 GB | 14.4 GB | RTX 4060 Ti 16GB |
| IQ3_XS | 11.5 GB | 15.3 GB | RTX 3090 24GB |
| IQ3_S | 12.0 GB | 15.9 GB | RTX 3090 24GB |
| Q3_K_XL | 13.1 GB | 17.3 GB | RTX 3090 24GB |
| IQ4_XS | 14.3 GB | 18.6 GB | RTX 3090 24GB |
| Q4_K_M | 16.5 GB | 21.3 GB | RTX 3090 24GB |
| Q4_K_XL | 17.6 GB | 22.6 GB | RTX 5090 32GB |
KV cache uses a flat +20% context/overhead rule because this repo does not publish its config; exact numbers may differ for long contexts. MoE models keep all experts in memory — total size counts, not just active params. Longer context grows the KV cache — use the selector above.
Can I run Qwen3.8-27B-AP on my GPU? (IQ2_S recommended quant)
| Hardware | Memory | Verdict | Est. speed |
|---|---|---|---|
| RTX 3060 12GB | 12 GB | no | |
| RTX 4060 Ti 16GB | 16 GB | RUNS | ~22 tok/s |
| RTX 3090 24GB | 24 GB | RUNS | ~73 tok/s |
| RTX 4090 24GB | 24 GB | RUNS | ~79 tok/s |
| RTX 5090 32GB | 32 GB | RUNS | ~140 tok/s |
| RTX PRO 6000 96GB | 96 GB | RUNS | ~140 tok/s |
| Mac 16GB unified | 16 GB | RUNS | ~8 tok/s |
| Mac 32GB unified | 32 GB | RUNS | ~12 tok/s |
| Mac 64GB unified | 64 GB | RUNS | ~21 tok/s |
| Mac 128GB unified | 128 GB | RUNS | ~31 tok/s |
| Mac 256GB unified | 256 GB | RUNS | ~43 tok/s |
| Mac 512GB unified | 512 GB | RUNS | ~64 tok/s |
| 32GB system RAM (CPU) | 32 GB | RUNS | ~4 tok/s |
| 64GB system RAM (CPU) | 64 GB | RUNS | ~4 tok/s |
| 128GB system RAM (CPU) | 128 GB | RUNS | ~6 tok/s |
| 256GB system RAM (CPU) | 256 GB | RUNS | ~6 tok/s |
| 1TB server RAM (CPU) | 1024 GB | RUNS | ~16 tok/s |
| 2TB server RAM (CPU) | 2048 GB | RUNS | ~16 tok/s |
Speed = decode tok/s estimated from memory bandwidth ÷ measured file size (× MoE active share). Real numbers vary ±30% by runtime and settings.
How much VRAM does Qwen3.8-27B-AP need?
The honest answer: 13.0 GB at the IQ2_S quant with 8K context — that is the measured
file size (9.6 GB) plus KV cache and runtime overhead. The smallest published build needs
13.0 GB; the lossless one needs 22.6 GB.
Cheapest hardware that runs it: New RTX 4060 Ti 16GB (~$420, used market) — see which GPU to buy.
Can I run it with CPU offload?
Yes, if combined RAM+VRAM ≥ file size — but generation speed is limited by memory bandwidth. Expect roughly: DDR5 dual channel ~10-30 tok/s for small MoE actives, single digits for big ones, 0.1-1 tok/s when experts page from disk. CPU-only is fine for batch jobs, painful for chat.
Why do other sites show different numbers?
Most pages compute weights from a formula (params × bits/8). We use the actual file sizes published in GGUF repos, which include the embedding table, unquantized tensors, and container overhead — that is why our numbers can differ from a naive calculation.
FAQ
How much VRAM does Qwen3.8-27B-AP need?
The measured answer: 13.0 GB at the IQ2_S quant with 8K context. The smallest published build needs 13.0 GB; the lossless (F16/BF16) build needs 22.6 GB. These are real GGUF file sizes from Hugging Face, not formula estimates.
Can I run Qwen3.8-27B-AP on a 24GB GPU (RTX 3090/4090)?
Yes — at IQ2_S (13.0 GB). Quants that fit 24GB: IQ2_S, IQ3_XXS, IQ3_XS, IQ3_S, Q3_K_XL, IQ4_XS, Q4_K_M, Q4_K_XL.
How fast will Qwen3.8-27B-AP run?
Decode speed is memory-bandwidth-bound. At IQ2_S: roughly 79 tok/s on an RTX 4090, 73 tok/s on an RTX 3090, 31 tok/s on an M-series Mac with 128GB, and 3.9 tok/s on CPU with dual-channel DDR4 — estimates from measured file size, MoE active-parameter share and memory bandwidth.
Can I run Qwen3.8-27B-AP on CPU with system RAM?
Yes, if your usable RAM is at least the file size. At IQ2_S you need ~13.0 GB of RAM. CPU generation is memory-bandwidth-bound: expect single-digit tok/s for large MoE models, 10-30 tok/s for small active params, and below 1 tok/s when experts page from disk.
Why do VRAM numbers for Qwen3.8-27B-AP differ between sites?
Most sites compute weights from a formula (params x bits/8). ModelFit uses the actual GGUF file sizes published on Hugging Face, which include the embedding table, unquantized tensors, and container overhead — so our numbers reflect what you really download and load.