Back
RCreddit.com
15
·17 hr ago·Dev community · RSS

Qwen 3.8 27b (Q4KM) oneshot a Super Mario clone

View original
LlamaQwenDeepSeekModel release

Heat trend

Collecting trend data

The percentage is based on available heat signal, not comment count or independent people.

Why it matters

Llama model activity is surfacing — worth tracking for capability changes, ecosystem impact, and availability.

AI summary

A developer was "blown away" by Qwen 3.8 27b (Q4KM) managing to "oneshot a Super Mario clone" despite their "crap" setup. The software used included LLAMA.cpp (Q4KM, xhigh, 8bit KV, MTP=1), Lmstudio Qwen 3.8 27b (Q4KM) GGUF, and Deepseek harness (mode: minimal). This achievement was considered "unbelievable" by the developer, highlighting the model's impressive capability in a single take.

I am absolutely blown away. Yes my setup is crap but the fact that it managed to do this in a single take is unbelievable (and I'm a developer).

Hardware used:
 - Windows PC with 4070ti (12GB VRAM, 32GB RAM)
 - Macbook M5 Air (LLAMA.cpp RPC connection to Windows PC)
Software used:
 - LLAMA.cpp (Q4KM, xhigh, 8bit KV, MTP=1)
 - Lmstudio Qwen 3.8 27b (Q4KM) GGUF
 - Deepseek harness (mode: minimal)

Prompt: "please create a fully self-contained super mario game with only one short level, put everything inside mario.html inside the current directory"

context: 64k thinking: xhigh time took: 117 minutes avg tps: 7.6

resut: https://pastebin.com/qyBu64sP

https://reddit.com/link/1w4821c/video/qpukeg1y4wmh1/player

Qwen 3.8 27b (Q4KM) oneshot a Super Mario clone · BuzzRadr