The week the answer stopped being a paragraph

Hello,

If you only take one thing from this week, take this: the answer is no longer a block of text. OpenAI shipped GPT-6 with Intelligent UI to everyone, which means ChatGPT now fills the canvas with visual and interactive components rather than prose. Anthropic, meanwhile, pushed in the other direction on economics, with Claude Haiku 5.5 landing at roughly 75% less to run than its predecessor, plus Motion and Dashboards in beta. One camp is reshaping what a result looks like, the other is making it cheap enough to generate at volume. Both change the maths on how discovery works for your brand.

1. GPT-6 and Intelligent UI roll out to everyone

OpenAI is rolling GPT-6 out globally in ChatGPT alongside Intelligent UI, which delivers faster responses with visuals and interactive experiences users can explore and use directly. As Michelle Pokrass put it, it is "a bet on model intelligence", letting GPT-6 fill the canvas with high utility visual and interactive components. The team had to build tools so those interfaces are token efficient, stream in instantly, match the design language, and stay expressive enough for anything a user might ask.

OpenAI's own framing is that everyday questions become more visual, complex topics become easier to grasp, and interactive tools for your task are available on the spot.

What it means for you: If the model can build a comparison tool, a calculator or an interactive explainer on the spot, a meaningful slice of the content you publish to capture that intent is now competing with a generated interface rather than another blue link. Audit your top-of-funnel assets and ask which ones exist only because the format was convenient. Those are the vulnerable ones. The durable ones are the pieces carrying proprietary data, original research, pricing, and lived expertise that a model cannot synthesise on the fly.

2. Claude Haiku 5.5 makes high-volume AI work genuinely cheap

Anthropic calls Haiku 5.5 the cheapest, fastest and most capable small model it has ever released, costing around 75% less to run on average than Haiku 4.5. Per token it is 90% cheaper on tasks under 100,000 tokens, which account for roughly 90% of requests to the previous Haiku model, and half the price on longer tasks.

It is built for high-volume, cost-sensitive work: summaries, classification, first-pass document reads, live support chat, routine computer use, and sub-agent duty alongside Opus 5.5 and Sonnet 5.5 on coding jobs. It is also the first Haiku with an adjustable effort setting, so you can tune for cost or intelligence per task. It is available now on all platforms including AWS, Google Cloud and Microsoft Azure. Anthropic is also halving cache read prices on Sonnet 5.5, making it around 20% cheaper on most long-running work.

What it means for you: The projects your team shelved last year on unit economics are worth reopening. Site-wide metadata generation, internal link analysis, bulk content classification, query clustering across hundreds of thousands of search terms, support deflection, these all become defensible at this price point. Pair a small model for the volume work with a larger one for judgement calls, and you get a workflow that scales without the invoice scaling with it.

3. Claude Motion and Dashboards enter beta, Docs and Slides go free

Claude Motion turns a report or chart into a short animated explainer, written as code rather than generated by a video model, so you can change any word, number or timing, then export an MP4 or open it in your video editor. It is in beta on Team and Enterprise plans. Claude Dashboards connects to your data platform or CRM and builds dashboards that update as your data changes, in beta on all paid plans.

Separately, Claude Docs, Slides and Design are out of beta and available on every plan including Free, with collaborative editing between your team and Claude on the same file.

What it means for you: Two things. First, the reporting layer of marketing work is being commoditised, and the monthly performance deck is a prime candidate for automation. Spend the reclaimed hours on interpretation, not assembly. Second, the code-not-video approach matters for brand teams, because editable output means a wrong figure is a text change rather than a re-render.

4. Ultrafast arrives for GPT-6.1 Sol

OpenAI is rolling out Ultrafast for GPT-6.1 Sol across the API, Codex and ChatGPT Work, promising near-Astra intelligence at up to 8x faster speeds than Sol Standard.

What it means for you: Latency has been the quiet blocker on customer-facing AI. At 8x, real-time use cases that felt clunky, on-site assistants, live personalisation, instant search refinement, move into the plausible column. If you benchmarked these six months ago and walked away, the test is worth rerunning.

5. ChatGPT for Teens and College Planner

OpenAI shared progress on ChatGPT for Teens, an under-18 experience that applies automatically to accounts identified as belonging to someone under 18, with protections on by default. Alongside it comes a preview of College Planner, which brings application requirements, deadlines, tasks and financial-aid steps together for US high school students planning a four-year college. New study tools and support for college advisers and teen voices are part of the same push, and OpenAI says it will keep evaluating how the protections work in practice.

What it means for you: If you market in education, youth products or anything family-adjacent, a segmented experience with default protections changes how your audience encounters you inside the assistant. More broadly, this is a signal about where the platform is heading: not one undifferentiated chat surface, but distinct experiences with distinct rules. Audience-specific AI visibility is about to become a real line item.

The thread

Intelligence is getting faster and cheaper at the same time that the output is getting richer. That combination tends to push value away from producing information and towards owning something the model cannot generate: your data, your customers, your point of view.

Next week I will be watching how publishers respond as Intelligent UI reaches scale, because that is where the first real traffic signals will show up.