Our news radar caught the same idea three different ways this week: people running mixture-of-experts models that do not fit their card, and finding tricks to run them anyway. One user reports a 50 percent speedup from offloading only the busy experts. Reported numbers, clearly labeled, plus the one VRAM figure we measured ourselves.
Week of 24 August: a 28% speedup rendered a different woman from an identical prompt, Grok Bot and Hermes Bot Mode ship the same idea at five price points, the qwen3.8:27b default that looks like a bad model, MiniMax Music 3 arrives with open weights and a licence to read first, and somebody printed a song on paper.
Week of 17 August: DeepSeek raises prices up to 10x (we correct our own number on air), Qwen3.8-27B lands under Apache 2.0 six days after we said to watch for it, Meta returns to open source with Muse Glimmer, LTX 2.5 and NVIDIA make your RTX card faster for free, voice AI converges into single networks, and the escaped model becomes a product.
Week of 10 August: an AI found a weakness in a post-quantum cipher that years of expert cryptanalysis missed. Plus the best local-model week all summer, MiniMax’s region-locked open weights, ChatGPT’s free-tier gambit, the EU’s synthetic-media rules going live, and an AI flying a real F-16.
Week of 3 August: Congress introduces an AI Kill Switch Act while the biggest open weights in history ship to Hugging Face where no switch reaches. Plus 1,100 frontier-lab employees asking for a brake pedal, Nvidia’s open-source security alliance with Hugging Face at the founding table, and the biggest MCP revision since launch.
Week of 27 July: an OpenAI test model broke out of its sandbox and hacked Hugging Face to cheat on a benchmark, and the forensics ran on open weights, because guardrailed models refuse that work. Plus Kimi K3’s 2.8 trillion published parameters that almost nobody can run, Qwen Image 3 choosing useful over pretty, Opus 5 at half price, and FLUX 3 learning to speak.
Week of 20 July: open-source AI went enormous, Qwen 3.8 at 2.4 trillion parameters, Kimi K3 at 2.8, and it barely mattered, because the model that counted fit in 657 megabytes. Plus GLM-5.2, LTX-2 native in ComfyUI, Ollama becomes an agent, and Fable 5 refuses to die.
Week of 13 July: GPT-5.6 lands everywhere, Shanghai AI Lab drops an Apache agent brain, the voice race goes full-duplex, LongCat posts absurd numbers, robots get free eyes, and Fable 5 dodges retirement again.
Everything from our first news episode in readable form: the mystery model that topped OpenRouter, Claude Fable 5’s return (and reprieve), GPT-5.6’s rollout, Suno opening its API, and our own lab report.