New: Framer Agent’s model picker now includes Reasoning and Fast Mode, so you can control how much a model thinks before building and, when supported, how fast it works. https://sub.chujiqiqi.click/updates/agent-reasoning-and-fast-mode
Introducing Video Shaders. Shaders in Framer can now accept both images and video, which we’ve grouped under Media in the Insert panel. We’re first updating the Chromatic Aberration shader with video support and all-new controls. Videos even play in realtime as you’re editing properties. With our new properties, you can add glass-like curvature and dispersion to your videos and images with a few clicks. High-fidelity effects, web-ready, published in seconds. https://sub.chujiqiqi.click/updates/video-shaders
GPT 5.6 Terra and Luna are 50% off through August 14.
Plus 32 Agent updates and 14 Canvas updates.
All in the July update. Now in Framer.
Opus 5 is now available in Framer. On our Navigation Benchmark, it completed the task 1.5 min faster than Opus 4.8, and ~12% more efficiently when creating fully interactive and responsive navigations from scratch. Opus 5 achieved the same score as Fable 5 while using 0.5× the credits in this task. It generally uses 1.2× the credits of GPT 5.6 Sol, our new baseline. https://sub.chujiqiqi.click/updates/opus-5
Fable 5 is now available in Framer. It’s the most proactive model we’ve tested. It goes beyond the brief, setting up styles and reusing elements across your site. When designing from scratch, first results are polished, with thoughtful finishing touches. It handles creative work like shaders and subtle animations noticeably better than models like Opus or Sonnet. In our latest evals, Fable 5 scored 83% overall, clearing Opus 4.8 at 77%, and led every model on design at 81%. At 2× the credit usage of GPT 5.5, Fable lands around 3.3× Sonnet 5. Now live in the agent model picker.
Sonnet 5 is now available in Framer. Our early use shows it’s a stronger partner for original design work than Sonnet 4.6, especially around layout, visual direction, and editing alongside you. It asks sharper design questions, uses images more purposefully when the prompt calls for it, and is better at picking fonts and writing in a more considered way. Expect a little more upfront thinking than 4.6, but strong efficiency in practice: in our Framer evals, Sonnet 5 scored higher overall while costing less than Sonnet 4.6, with accuracy improving from 72% to 90%. It uses 0.6× the credits of GPT 5.5, with introductory pricing offering 33% off through August 31.
Publish
Liquid Glass
We’ve been heads-down for a long time, working on an all-new, completely redesigned Framer. I’m very proud of how it turned out. Framer 3.0 is coming soon.
