Okay. Wildddddd!! Grok 4.5 beat Kimi K3 on the exact same task for 13x cheaper 👀
Aimlapi ran the test: same prompt, same one-shot job, no tricks Cost per figure: Grok 4.5 → $0.15 GPT 5.6 Sol → $0.60 Qwen 3.8 Max → $0.67 Kimi K3 →…
Qwen 3.8: There's a free 2.4 trillion parameter AI that goes head-to-head with the biggest names.
It just landed, and most people have no idea it exists. → Reads text, images, and video → Holds 1M tokens of context → Works on projects for days…
Qwen 3.8 Max Full COURSE 1 HOUR (Build & Automate Anything)
The biggest upgrade isn't another model.
It's shared memory. Every Hermes conversation gets written into an Obsidian vault.
Zuckerberg is finally building AI.
China ran out of GPUs. Microsoft is losing relevance. The American AI monopoly just ended.
Hex-A-Gone from Fall Guys, rebuilt in Three.js.
Same prompt, 4 models: grok 4.5 - 10/10. smoothest movement of the four, visuals hold up under motion, and it cost less than opus.
Kiana vs Qwen 3.8 max
Photorealistic 3D luxury countryside house in Three js with authentic architecture, a landscaped garden, carport, glass sunroom, pergola, hot tub,…
DeepSeek V4 Flash AI is 100x cheaper than Claude Fable 5, Kimi K3, Qwen 3.8 Max, and Opus 5
A FREE AI just replaced $200 Microsoft Office, and Meta dropped its own Claude Code rival these are the 16 biggest AI updates this week, in one video.
The black art of lora training is starting to clear up for me a little bit finally had a pretty good poet training result.
Trying a "terminal" style for websites used Qwen 3.8 Max as the teacher, training Gemma 4 12b (because it's fast).
Qwen 3.8 Max asked it for a forest fire simulation but let it go crazy with different styles, and the sim rules.
Picked some of my favorites here. All look so varied and turned out incredibly well and fun to watch. All 0 shot.
Someone built A mystery Qwen model called Kiana just appeared on Arena.
Early “ballistic” tests are already putting it in the same conversation as Claude Opus 5 and GPT-5.6 Sol.
Alibaba just built an AI agent that beat GPT-5.6, gemini and claude at using a computer.
But the benchmark isn't even the wildest part. Qwen 3.8 Max: → 2.4 trillion total parameters → 95B active parameters per task → 1 MILLION token…
Is open source AI actually winning in 2026, or is that just a headline?
Five major models say a lot more than you'd expect. Kimi K3, Qwen 3.8, GLM 5.2, Ornith 1.0, and DeepSeek V4 Pro all shipped within about four months…
Kimi K3 won't just recreate this.
It built the whole thing from scratch. A rooftop chase across a fully lit city at night.
Day 13 Implemented the Battle Pass
We now have actual progression and a reason to keep coming back Opus 5 orchestrated Qwen 3.8 Max, GPT-5.6 Luna Max and Grok 4.5 to get it done We…
Qwen3.8-Max vs Kiana (unreleased Qwen model)
A new Qwen model "Kiana" spotted in arena it faster than qwen3.8-max but not as good as the max model looks more like Qwen3.8-27b open model which is…
AI models like Qwen 3.8 max have demonstrated the capability to run multiple e-commerce stores for a…
AI models like Qwen 3.8 max have demonstrated the capability to run multiple e-commerce stores for a full year, quadrupling their initial capital and…
Stop using AI like 5 employees locked in 5 different buildings.
I built a Hermes Agent OS that gives my AI agents one screen, one shared memory, and the ability to work for hours without me.