BREAKING: @OpenAI just dropped GPT-6 Sol. It’s my new daily driver in Codex: not quite Astra, but close enough for much of my everyday work, faster, and 50% cheaper than 5.6 Sol. We tested it across the work we actually do at @every and ran it head to head versus Opus 5.5. Here’s my vibe check: • Writing: On a paragraph-writing task drawn from my real work, Sol scored close to Astra, which is still my top model for writing. It writes clean, minimal prose and puts the important idea first. Opus 5.5 is pleasant to work with, but its drafts still tend to bury the point. • Computer use: If you love Astra’s computer use, you’ll like Sol. An earlier Sol preview matched Astra on 17 of 18 attempts across six of our simpler Hands tasks, at a much lower token price. • Coding: Sol improves on GPT-5.6—including better Ruby code in @kieranklaassen tests—but Opus 5.5 has the higher ceiling for long, autonomous builds. One frustration: Codex’s new security classifier repeatedly stopped work we’d already authorized to ask for approval. That’s friction from the classifier, not necessarily a limitation of Sol, but it made long runs harder to leave alone. Overall: Sol feels like an S-class iPhone release: It will give you much of Astra’s power at about a fifth of Astra’s price. Opus 5.5 is the bigger surprise. In some of our tests, it matches or even beats Fable 5.1 on many tasks while also being cheaper than Opus 5. A few people on our team who were Codex converts have started to wobble with Opus 5.5. If you’re already in the Claude ecosystem you’re going to love this model. If you’re a Codex user, it’s worth a look especially for your top-end coding tasks. You’ll like Sol 6 if you spend your day in Codex reading, writing, and getting things done. It’s fast, much cheaper than Astra, and it’s the model I keep reaching for. You’ll like Opus 5.5 if you want to hand an agent a hard coding or visual project and see how far it can take it. Its best work went further than Sol’s in our tests—enough to pull some of our team back toward Claude. You can go deeper on all of our evals including each real-world task and how they're scored at the links below: Writing: Knowledge work: Reading: Full vibe checks on @every coming soon!

Working with AI is frictionless. One message produces polished landing pages, complete brand guidelines, and working apps. But we are skipping the “productive friction,” says @vanlancker, partner at @terraincap. Learning happens through trial and error: Attempt, fail, dissect, try again. Preserving friction, Willem argues, could make work in the AI age more methodical and meaningful. Read Willem’s Thesis Statement:

any friends willing to nominate me for Every’s Thesis: 2027? AI has helped me go from designing products to building them. I’d love to compare notes with other builders and bring what I learn back to the Southeast.

AI is getting good at video generation. But the movie industry won’t die. Human actors and screenwriters aren’t going anywhere, because acting and screenwriting are jobs only humans can do well. @NirZicherman, cofounder and CEO of @oboelabs, expects models to streamline financing, casting calls, and logistics—work peripheral to the creative core. Read Nir’s Thesis Statement:

.@kplikethebird on building an AI writing plugin that improves over time. Katie doesn't write in a chat window anymore. Instead of starting fresh in a Claude or ChatGPT project, every piece she writes runs through a system she built in the Claude desktop app—that already knows her voice, her style, and how she wants a draft to come together. That system is Compound Writing. In February, she downloaded @kieranklaassen's Compound Engineering plugin and asked Claude to help her build the same thing, but for writing. She's spent the past six months building on it since, drawing on more than a decade of professional writing experience. In a recent AI & I episode, Katie unpacks how she built the plugin and how she uses it day-to-day. Link to the full episode:

A year ago today, we launched monologue There’s so much we want to build, and I’m excited to keep working on it To everyone who’s used it, shared it, or told us how to make it better: thank you! Here’s to year two 🚀

The more I think about this, the more @every looks like a preview of how product development might work when software is cheap. Their products often start as something they need themselves. If it spreads across the team, they put it in front of readers. Cora, Sparkle, Spiral and Monologue all came out of that loop. The product becomes a way to find out whether the same problem shows up elsewhere too. I wrote about why I think more of us can work this way now:

Huh. An AI that can’t write text and can only predict things?? Stochastic parrot is out Smart If statements are in? Watch the DOOM demo they buried in the thread. Super cool. Arbitrary predictions feels way closer conceptually to actual brains than LLms imo Not a text engine A prediction thinking engine

Huh. An AI that can’t write text and can only predict things?? Stochastic parrot is out Smart If statements are in? Thinking engine, not text engine Watch the DOOM demo they buried in the thread. Super cool. Arbitrary predictions feels way closer to actual brains and AI than LLms imo