And just like that... we're off! So are the markets 😃 $NVDA and $CRWD leading the charge — two of my faves lighting up the tape today. Told ya the AI trade wasn't done. 💯 #SixthWave#

$PLTR's AIP Bootcamps are now routinely delivering production use cases in 5 days. Sovereignty Bootcamps just pulled in nearly 200 organizations in under 30 days. Defense primes and enterprises are no longer piloting for months — they’re embedding AI at the acquisition and architecture stage. The demand signal is unmistakable: decision velocity commands a premium. Faster proof-of-concept → broader buy-in → platform lock-in → margin expansion as usage scales. We're seeing the Sixth Wave in action. Software accelerating the world — and the firms that move first are pricing decades of institutional advantage into a single fiscal year. And we're still in early innings.

The consensus for some time has been that we are getting a new Anthropic model ahead of the IPO (and that's not all), so I think we can safely say this is 100%. There were some new developments today from @synthwavedd though, who says that new Sonnet, Opus, and Fable are all in stealth testing. In fact, if you are using Claude, it's possible that you are in an early test flight and using one of them right now. So we may be getting a triple release from Anthropic, not just one new model.

Ever ask @ChatGPT to fix one tiny detail in an image and get back a whole new picture? OpenAI's GPT Image 2.6 guide has some practical tips for getting the image you want and keeping the parts you already like. 1. Say what you're making. A wide launch banner, a portrait, and a product photo need different layouts. Give the model the format and say where the main subject should go. 2. Describe what you'd actually see. "A rocket on the right, pink light on the exhaust, and dark sky on the left for a headline" gives the model more direction than "epic synthwave rocket." 3. Put exact text in quotes. Try "Write 'Omarchy 4.5' once at the top in large white letters." Then zoom in and check the spelling and legibility before posting. 4. Edit one thing at a time. Use the last good image and name what should stay fixed. "Remove the letter from the rocket. Keep the logo above it, the colors, and the framing." If a logo needs to be pixel-perfect, paste the original back in after the edit. 5. Give every reference image a job. Tell the model to use image 1 for the layout, image 2 for the logo, and image 3 for the color palette. That makes it clear what to take from each one. 6. For API edits, circle the target. GPT Image 2.6 accepts a transparent markup layer, so you can point to the exact object you want changed and ask for the circle to disappear in the final image. 7. Tune planning and detail separately in the API. Try low complexity for a simple edit. If the layout looks right but the small print is blurry, raise quality. Start with Flare when speed matters, and try Sunburst for demanding layouts or precise edits.