Happy weekend AI believers
A lot of cool AI models have been released recently -> GLM-5.3 -> DeepSeek V4 Pro 0813 -> Gemini 3.7 Flash -> Grok 4.6 -> GPT-5.6 Cyber -> Muse Spark 1.2 -> Qwen3.8-12B which one impressed you the most?
The harness multiplier: same AI.
Same Tasks. 20-Point Swing. A public test just proved something huge: Your SETUP is quietly deciding half your AI results. The test: one model — DeepSeek V4 Flash. Eight different harnesses. 30 hard agentic tasks. The scores: Pi Agent: 20/30
A 4090 running through the full stack of AI video calls: Qwen3.8-27B local deployment, conversation delay of 2 seconds
Qwen3.8-27B has just been open sourced. I spent one night deploying it to the local 4090. I changed the LLM brain of the self-developed open source framework VoxEMW from the DeepSeek API to this local 27B. The measured delay dropped from 2.4 seconds to 2.0 seconds.
Kimi K3 seems to be cooked for me, and honestly, Grok 4.6, Gemini 3.7, DeepSeek V4 are way better.
Built Football Frenzy, a playable mini soccer game demo, and saved the version (ID: db774d2). What's in the game: • 5-a-side match on a striped canvas pitch with full
You have reached the end of the archive
All of deepseek