OpenAI CEO Sam Altman laid out the company’s ChatGPT plans for the near future. After plenty of speculation and rumors about the next big ChatGPT upgrades, we know that GPT-4.5 and GPT-5.0 models will be available to users in the coming weeks. But it’s not just OpenAI that is about to drop big AI upgrades.
Anthropic CEO Dario Amodei said in an interview about a month ago that Claude will receive a reasoning model similar to ChatGPT’s o1 and o3. Also, Claude will catch up to ChatGPT in another big way, as it’ll support internet searches. Amodei did not specify timelines for either upgrade, but it looks like it’s all coming soon.
Anthropic has yet to announce the new Claude features, but someone poked through a recent update for the iPhone app and found a few telling clues. The Claude app includes new icons that seem to confirm the latest features. They also suggest Anthropic might give Claude a big upgrade, which people call Claude 4 online.
An X user named M1Astra shared the following findings in the Caude iOS update:
BREAKING: Anthropic set to release Thinking Model/Mode & Web Search capabilities imminently, based on findings in the latest update to the Claude iOS app.
New symbols like “Steps” & “Think” suggest a thinking model, while “MagnifyingGlass” & “Quotes” points to web search.
“Nodes-Think,” “ClockCountdown,” “Lightbulb,” “Gauge,” “Steps,” “MagnifyingGlass,” “Quotes”
The user also shared images that show new icons Anthropic supposedly added to the Claude iPhone app.
These assumptions make perfect sense considering what OpenAI did with the ChatGPT composer. The Composer is the section of the ChatGPT UI where you enter commands for the AI. OpenAI updated it over the years with new icons that act like shortcuts to new features. For example, you can tap a button to use ChatGPT Search with your prompt.
The ChatGPT composer also has buttons for Advanced Voice Mode and Canvas. It also lets you upload files and generate images.
The model picker sits at the top of the chat. You can switch between the reasoning models (o1, o3-mini, and o3-mini-high) and regular ChatGPT models like GPT-4o.
Given the findings in the iOS app code, it’s likely Anthropic will follow a similar route. Adding more icons to Claude’s Composer will make it easier to use the new features. As for the reasoning model, it’ll be interesting to see how Anthropic implements it.
I’ll also point out that OpenAI’s o3-mini models are the latest reasoning upgrades for ChatGPT. In addition to offering better performance than o1, they also support internet search, which is a great tool to have on hand when asking the AI questions.
While Anthropic might be matching OpenAI’s reasoning models and internet search support, there’s no telling whether Anthropic will also make any AI agent announcements soon. Technically, Anthropic unveiled an AI agent before ChatGPT got Operator and Deep Research, though OpenAI’s approach to agents seems better for the time being.
There’s no telling when Anthropic will unveil its Claude 4 upgrade. But given the code findings, it’s likely some sort of announcement will be made soon.
