📫 Subscribe | 🗣️ Hire us to speak | 🎓 Online Courses | 📅 Events

The red card shown to Anthropic’s Fable 5 and Mythos 5 models by the federal government in mid-June was rescinded on July 1; in non-soccer terms, this means that the two models that had previously been banned are once again available to the public. There’s one caveat, though: starting on July 7 (i.e., today), access to Fable 5 will only be via usage credits, which means that Claude subscribers will no longer have access to the model without incurring additional charges. Mythos 5, the cybersecurity-oriented model that was never released to the public to begin with, will be made available to a select group of US organizations, which is the same arrangement in place before the model was subjected to export controls by the US government.

Last Tuesday, Anthropic launched yet another Claude model, updating their middle-tier Sonnet model to version 5 from version 4.6. Why does this matter? Across a number of benchmarks, Sonnet 5 nearly matches the performance of the larger Opus 4.8, particularly in agentic tasks, continuing Anthropic’s trend of attaining higher levels of performance at a lower cost (in this case, 40% less per token compared to Opus 4.8). This trend of producing smarter but cheaper models can not only yield potential savings for organizations that rely heavily on Claude models but also strengthens Anthropic’s IPO narrative on the company’s path to going public.

Last week Google launched Nano Banana 2 Lite, an image generation model that is considerably faster than the non-Lite Nano Banana 2: “You can go from text to an image in about 4 seconds in the default low-thinking mode. Generating the same images in the standard Nano Banana takes about 20 seconds.” Unlike the trend of cheaper and smarter language discussed above, the speed offered by Nano Banana 2 Lite comes with tradeoffs, as the model struggles with text embedded in images as well as keeping characters consistent across images. In addition, infographics produced by the model are more likely to include incorrect data.

Lean TECHniques works alongside Midwest companies to leverage technology for business impact. Through advisory, coaching, and embedded delivery, Lean TECHniques helps organizations develop high-performing teams, create environments that fuel innovation, and deliver best-in-class customer experiences. Core areas of expertise include software, cloud, artificial intelligence, product management, and technology leadership.

NotebookLM (one of our favorite AI tools), recently launched Short Video Overviews. Find them in the Studio tab by pressing the arrow next to “Video Overview.” This turns your uploaded docs into 60-second narrated vertical videos. This is great for turning dense training material into something people will actually finish.

Testing out another one of ChatGPT’s pre-written image prompts (with a few modifications).

Microsoft is developing its own AI models, and you can test them out. The MAI (Microsoft AI) Playground is a free, web-based demonstration platform where users can test and experiment with Microsoft's in-house artificial intelligence models. It allows developers, creatives, and consumers to try experimental models directly before they are fully deployed across commercial products.