Claude Platform adds Opus advisor as OpenAI launches $100 ChatGPT Pro tier

Anthropic introduced its “advisor strategy” for the Claude Platform on April 9. Sonnet or Haiku handles a request as the executor and can consult Opus through a server-side tool. The exchange stays within a single Messages API request. The feature is available in beta on the Claude Platform for developers building applications with Claude.¹

The approach separates the work of completing a task from occasional consultation: Opus does not have to handle the entire request.¹ In Anthropic’s own tests, Sonnet with Opus as an advisor performed better than Sonnet alone while reducing the cost per task.² Those results do not establish whether the same benefit applies to other workloads. Teams will need to measure quality and costs against their own use cases.

OpenAI introduced a $100-per-month ChatGPT Pro tier on April 9, 2026. It is intended for people using Codex in longer, demanding coding sessions. Its standard Codex allowance is five times that of ChatGPT Plus.¹² The new tier adds to OpenAI’s existing subscriptions rather than replacing the more expensive Pro option.³⁴

The plan sits between Plus and the existing Pro tier on price.³⁴ It arrives as OpenAI and Anthropic compete for users of AI coding tools.³ When comparing plans, the lasting allowance matters more than the launch offer: OpenAI is temporarily providing more Codex usage, after which the standard fivefold allowance relative to Plus applies.¹ Whether that is enough for a team depends on its actual workload.

Google announced “Notebooks” in Gemini on April 8, 2026.¹ Users can group chats and files for a project, add sources, and set instructions.¹ Notebooks sync with NotebookLM, so sources added in either app are available in the other.¹ The rollout starts on the web for Google AI Ultra, Pro, and Plus subscribers.¹

The feature connects project work in Gemini with source-based work in NotebookLM.¹ NotebookLM remains available for further tasks, such as turning sources into other formats.¹ Google says mobile access, more European countries, and free users will follow in the coming weeks, but it has not given an exact date.¹ For now, availability depends on the user’s subscription and whether the rollout has reached their account.¹

Amazon CEO Andy Jassy disclosed a revenue figure for AWS’s AI business for the first time in his April 9, 2026, shareholder letter. Its annualized revenue run rate exceeded $15 billion in the first quarter.¹² Jassy put the annualized run rate of Amazon’s chip business at more than $20 billion.²

The figures show the scale of both businesses within AWS: Jassy put the cloud division’s overall annualized revenue run rate at $142 billion.² Amazon plans about $200 billion in capital expenditures in 2026 and says customer commitments cover a substantial share of the planned spending.² These run rates extrapolate current performance over a year; they are not revenue earned over a full year or separately audited segment results.¹²

Florida Attorney General James Uthmeier announced an investigation into OpenAI on April 9. Among the issues under examination is an allegation that the gunman in the Florida State University shooting may have used ChatGPT to plan the attack. The inquiry also raises broader public safety and national security concerns about the service.¹²

ChatGPT’s alleged role in planning the shooting has not been proven. The inquiry puts attention on how an AI provider handles possible signs of serious violence and cooperates with authorities. For companies that provide or integrate AI assistants, the case highlights a practical question: safeguards depend not only on what a model says, but also on how risk policies are applied when concerning use is identified.¹³