If you use ChatGPT, Claude, or Gemini at work, the next three months are going to feel like trying to keep up with a software update every other week.
New models launching. Old models disappearing. Prices shifting. Features changing. It’s a lot — and most of the coverage is written for developers and AI researchers, not the rest of us who just want to know whether the tool we use every day is about to get better, worse, or more expensive.
So here’s the plain-language calendar. Every major AI change coming between now and June, what it actually means, and whether you need to do anything about it.
The Big Picture: Why Q2 2026 Is Different
Most quarters bring one or two AI updates worth paying attention to. This quarter has at least six.
OpenAI and Anthropic are both preparing to launch their most powerful models ever — possibly within weeks of each other. Google has a major conference with expected announcements. Microsoft is rolling out a new AI management platform. And several older models are being retired, which could break tools you rely on.
The sheer volume is unusual. As one AI observer put it: “It’s going to be a crazy month.”
Here’s everything on the calendar, sorted by date.
April 6-9: HumanX Conference (San Francisco)
What it is: A major AI conference with speakers from AWS, Stanford, and Databricks. Think of it as a place where companies preview what’s coming next.
Should you care? Only if you follow AI news closely. Any big announcements will be covered in the press within hours. You don’t need to watch it live — but if something major drops here, we’ll cover it.
Action needed: None. Just watch for news.
April 19: Claude Haiku 3 Goes Away
What it is: Anthropic is retiring Claude Haiku 3, one of their fastest and cheapest AI models. If you use Claude through an app, a chatbot, or a tool that runs on Haiku 3 behind the scenes, the model powering it will change.
What is Haiku? Claude comes in three sizes: Haiku (fast and cheap), Sonnet (balanced), and Opus (most capable). Haiku 3 is the older fast model. Anthropic is replacing it with Haiku 4.5, which is faster and smarter — but it works slightly differently.
Who’s affected:
- If you use Claude through claude.ai (the website or app): You won’t notice. Anthropic has already switched you to newer models.
- If you use a business tool that runs on Claude: Your tool provider needs to update. If they don’t, the tool may stop working or behave differently after April 19.
- If you built something using the Claude API: You need to update your model ID from
claude-3-haiku-20240307toclaude-haiku-4-5-20251001before April 19.
Action needed: If you use Claude through a third-party tool (like a customer service bot or content tool), ask your provider if they’ve updated to Haiku 4.5. If you’re a developer, Anthropic’s migration guide walks you through the changes.
April 22-24: Google Cloud Next
What it is: Google’s biggest cloud and AI conference of the year. This is where Google typically announces new Gemini models, AI features for Google Workspace, and developer tools.
What to expect:
- Updates to Gemini 3.1 Pro — Google’s most capable model, which competes with Claude Opus and GPT-5.4
- New features for Gemini in Google Workspace — smarter AI in Docs, Sheets, Slides, and Gmail
- Possible pricing changes for Gemini API access
- New tools for building AI agents (Google’s answer to Microsoft Copilot)
Should you care? If you use Google Workspace at work, yes. Any AI improvements to Docs, Sheets, and Gmail directly affect your daily workflow.
Action needed: None yet. Watch for announcements April 22-24 and we’ll cover what matters.
April 30: Claude 1M Context Window Beta Ends
What it is: Anthropic has been testing a “1 million token context window” — which basically means Claude can read and analyze extremely long documents (entire books, huge codebases, months of emails) in a single conversation. The beta period ends April 30.
What happens next: The 1M context window is moving from beta to the standard Claude Sonnet 4.6 and Opus 4.6 models. If you’ve been using the beta, you’ll need to switch to the regular models (which now include the same capability).
Who’s affected: Mostly developers and businesses using the API. If you use claude.ai directly, this shouldn’t change anything for you.
Action needed: If you’re a developer using the 1M context beta, update to the standard model IDs before April 30.
May 1: Microsoft Agent 365 Launches
What it is: Microsoft is launching Agent 365, a new platform that lets companies manage and monitor AI agents across their organization. Think of it as a control panel for all the AI bots and assistants a company uses.
What it costs: $15 per user per month, on top of existing Microsoft 365 subscriptions.
Should you care? This is primarily for IT teams and managers, not individual users. But if your company uses Microsoft 365, you might start seeing new AI features and policies roll out after May 1. Your IT department will be the ones deciding whether to adopt this.
Action needed: None for individual users. If you’re a team lead or manager, ask your IT department whether Agent 365 is on their radar.
May 6: Code with Claude Conference (San Francisco)
What it is: Anthropic’s first-ever developer conference. This is where they’re expected to announce new Claude features, possibly including the much-rumored Claude Mythos model.
Should you care? Even if you’re not a developer, this conference could announce changes that affect everyone who uses Claude — new models, pricing changes, or new capabilities. If Mythos launches here, it could be a significant upgrade.
Action needed: None. We’ll cover any announcements that affect regular users.
Coming Soon (No Exact Date Yet)
OpenAI Spud (GPT-5.5)
What it is: OpenAI’s next major model, codenamed “Spud.” Pretraining is confirmed complete — it’s just being tested and refined before release. Likely launch: mid-April to May 2026.
What to expect: A significant jump in reasoning and problem-solving ability over GPT-5.4. Some early reports suggest improvements in math, science, and complex analysis.
Will ChatGPT Plus users get it automatically? Almost certainly, yes — OpenAI has always rolled new models into existing subscriptions. You might see it appear as a model option in ChatGPT, or it might replace GPT-5.4 as the default.
Price concern: There’s growing anxiety that each new model brings higher subscription prices. ChatGPT Pro already costs $250/month, and Claude Max is $200/month. Whether a Spud-tier plan costs more than the current $20/month Plus plan is unknown, but OpenAI’s recent $122 billion funding round (the largest in tech history) suggests they’re building toward premium tiers.
Claude Mythos
What it is: Anthropic’s most powerful model ever — a tier above the current Opus. Leaked internal documents describe it as having exceptional reasoning, coding, and cybersecurity capabilities.
When: Possibly at the Code with Claude conference on May 6. Some analysts say Q3 2026.
The pricing question: This is what’s making people nervous. Leaked blog drafts described Mythos as “very expensive to serve.” Rumors suggest API pricing of $100 per million input tokens and $500 per million output tokens — dramatically more than current models. Whether regular Claude Pro subscribers will get access, or whether Mythos will require a new premium tier, is unknown.
The bigger concern: As one product manager put it: “We’re moving into a three-tier reality. Utility models for basic stuff. Standard models for day-to-day. And Mythos-class models for high-stakes work where cost is secondary to accuracy.” The question isn’t whether AI is getting smarter — it’s whether the smartest AI is getting more expensive faster than it’s getting more accessible.
Grok 5
What it is: xAI’s next model, with 6 trillion parameters (the largest publicly announced AI model ever). Being trained on the Colossus 2 supercluster.
When: Q2 2026 expected, no specific date announced.
Should you care? Only if you use Grok (available through X/Twitter). For most professionals, ChatGPT and Claude are more relevant to daily work.
DeepSeek-V4
What it is: The next model from DeepSeek, a Chinese AI lab that has surprised the industry with models that compete with much larger Western labs at a fraction of the cost.
When: Expected Q2 2026.
Should you care? DeepSeek models tend to be cheaper and open-source. If pricing of premium models like Mythos becomes a concern, DeepSeek could become a more affordable alternative.
The Calendar at a Glance
| Date | What’s Happening | Who’s Affected | Action Needed |
|---|---|---|---|
| Apr 6-9 | HumanX Conference | AI watchers | None |
| Apr 19 | Claude Haiku 3 retires | Developers, tool providers | Check if your tools updated |
| Apr 22-24 | Google Cloud Next | Google Workspace users | Watch for Gmail/Docs AI updates |
| Apr 30 | Claude 1M context beta ends | API developers | Update model IDs |
| May 1 | Microsoft Agent 365 launches ($15/user/mo) | IT teams, managers | Ask IT if it’s on the roadmap |
| May 6 | Code with Claude Conference | Everyone who uses Claude | Watch for Mythos announcement |
| May 19 | Code with Claude (London) | European Claude users | Same |
| June 10 | Code with Claude (Tokyo) | Japanese Claude users | Same |
| TBD Apr-May | OpenAI Spud (GPT-5.5) launches | All ChatGPT users | Should auto-update |
| TBD Q2 | Claude Mythos launches | Claude users | Watch for pricing |
| TBD Q2 | Grok 5 launches | X/Twitter users | None unless you use Grok |
| TBD Q2 | DeepSeek-V4 launches | Budget-conscious AI users | Potential cheaper alternative |
What You Should Actually Do
If you’re reading this and feeling overwhelmed — don’t be. Here’s the short version:
If you use ChatGPT: Do nothing. When Spud launches, you’ll get it automatically. Your $20/month plan will keep working. Watch for announcements about pricing changes, but don’t stress until they happen.
If you use Claude: Make sure your apps and tools are updated before April 19 (Haiku retirement). Watch the May 6 conference for Mythos news. Your current Claude Pro plan isn’t going anywhere.
If you use Google Gemini: Watch Google Cloud Next (April 22-24) for Workspace AI updates. Any improvements to Gemini in Docs or Gmail will roll out automatically.
If you manage a team: Look into Microsoft Agent 365 if you’re on Microsoft 365. At $15/user/month, it’s a real cost — make sure the governance and security features justify it for your organization.
If you’re worried about AI pricing: You’re not alone. The trend toward more expensive premium models is real. But the models most people use daily ($20-30/month plans) aren’t going away. The premium tiers (Mythos, ChatGPT Pro) are for businesses and power users, not the average professional.
The best thing you can do right now? Get better at using the tools you already have. A well-crafted prompt on today’s models will outperform a lazy prompt on tomorrow’s models every time.
Sources:
- AI Models in April 2026: Every Major Release — RenovateQR
- Claude Model Deprecations — Anthropic
- Claude Migration Guide — Anthropic
- Microsoft Agent 365 GA May 1, 2026 — Microsoft Community
- Agent 365: Enterprise AI Control Plane — Rob Quickenden
- Grok 5: 6T Parameters — NxCode
- OpenAI $122B Funding at $852B Valuation — Forbes
- Google Cloud Next 2026 — Egen
- AI Updates Today — LLM Stats