Builds / RTX 5060 Ti 16 GB
Single RTX 5060 Ti 16 GB: Qwen3.8 27B IQ4_XS at 27.6 tok/s, 50.2 with MTP (stock llama.cpp)
Source: github.com ↗📏 How measured: reported by the original source; RigBase re-checked the numbers at that source
- GPU(s)
- RTX 5060 Ti 16 GB
Amazon · eBay (used) · Newegg - VRAM / unified memory (GB)
- 16
- CPU
- not stated
- Motherboard
- not stated
- RAM
- 15 GiB system RAM (as reported)
Amazon · eBay (used) · Newegg - PSU
- not stated
- Other
- Stock llama.cpp built from source (commit 23b0202a); full GPU offload, 8192 context, q8_0 KV cache
- Reported total cost
- not stated
- Power
- not stated
Measured speed (generation tokens/sec)
| Model | Quant | Engine | tok/s | Prompt processing (tok/s) | Details |
|---|---|---|---|---|---|
| Qwen3.8 27B | Unsloth UD-IQ4_XS, q8_0 KV | llama.cpp 23b0202a, no speculative decoding | 27.6 | not stated | context: 8k ctx, 30-token prompt (26.5 tok/s at ~6.6K prompt)📏 Mean of 2 timed requests, 30-token prompt |
| Qwen3.8 27B | Unsloth UD-IQ4_XS + mtp-Qwen3.8-27B-Q4_0 head, q8_0 KV | llama.cpp 23b0202a, MTP-1 | 45.8 | not stated | ⚡ spec-decode (MTP)context: 8k ctx, 30-token prompt📏 Mean of 2 timed requests, 88.5% draft acceptance |
| Qwen3.8 27B | Unsloth UD-IQ4_XS + mtp-Qwen3.8-27B-Q4_0 head, q8_0 KV | llama.cpp 23b0202a, MTP-2 | 50.2 | not stated | ⚡ spec-decode (MTP)context: 8k ctx, 30-token prompt (48.1 tok/s at ~6.6K prompt)📏 Mean of 2 timed requests, 77.2% draft acceptance |
llama.cpp GitHub discussion by SoundByGAIA, 11 Oct 2026. No prompt-processing figure given. CPU, board and PSU not stated.
Source: https://github.com/ggml-org/llama.cpp/discussions/30328
Blank fields = not stated in the source. We never estimate build data; GPU street prices above are separately sourced and dated.
Rent before you buy: try this model on a cloud GPU for a few dollars first: Runpod (referral link; sign up with Google to get the new-user bonus)
Quick answers: Can RTX 5060 Ti 16GB run Qwen3.8 27B?