Qwen 3.8 27B
Deepseek Harness Time to complete: 13 1/2 minutes Average tokens per second: 369 RTX5090
0x Alpha had no business being this good at building games.
I asked four AI models to build a Mario-style game GPT-5.6 Sol Pro: 28:36 · 36.6K tokens · $0.72 0x Alpha Max: 52:25 · 86K tokens · FREE Qwen 3.8 Max: 15:43 · 29.5K tokens · $0.03 Grok 4.6 xhigh: 39:39 · 173K tokens ·
Qwen 3.8-27B just dropped.
Free, open weights, runs locally. It performs close to Claude Opus 4.6, the frontier model from just 6 months ago. No cloud. No subscription. Just your own machine. The catch: it needs real hardware. A Mac mini won't cut it. Set it up through Ollama
How I Ranked #1 in 7 Hours with Qwen 3.8 Max (FREE!)
Ranked #1 in 7 hours using Qwen 3.8 Max. ⚡ The proof: → Ranking above AI Overviews for "Hermes V0.2" in 7 hours → 0 → 224 clicks/day → 0 → 105 clicks/day Why Qwen 3.8 Max: free at (or API for
You have reached the end of the archive
All of qwen38