4/12 AI Express: GPT-4o Takes Full Charge / Genspark’s Rapid Revenue Generation / Google Cloud Goes All-In on AI Agents

News

  • 🤖GPT-4 Retires Next Year, GPT-4o Fully Takes Over ChatGPT
  • 🚀AI Agent Genspark Revenue Generation Myth: Achieved Ten Million Dollars ARR in 9 Days
  • 🔮OpenAI Hints at GPT-4.1: Million Token Context and Real-time Memory?
  • 🛠️Google Cloud Next Big Explosion: Agent Development Kit, AI Assistant, and New Model Chip Released Simultaneously

01

GPT-4 Retires Next Year, GPT-4o Fully Takes Over ChatGPT

OpenAI recently officially announced that its landmark large language model GPT-4 will cease service on the ChatGPT platform starting April 30, 2025. At that time, the superior-performing GPT-4o model, particularly excelling in writing, coding, and STEM fields, will fully take its place, becoming ChatGPT’s mainstay. Although GPT-4 is about to exit in front-end applications, OpenAI emphasized that developers and researchers can still access GPT-4 and other older versions of models through the API, ensuring the continuity of existing applications and research.

Core Highlights

  • Official Retirement Announcement: OpenAI clearly states that starting April 30, 2025, users will be unable to select GPT-4 service in ChatGPT, marking the end of its front-end application lifecycle.
  • New King Ascends: GPT-4o will completely replace GPT-4’s role in ChatGPT, with OpenAI stating it surpasses the former in multiple aspects, being the logical “natural successor.”
  • Interface Retention: Developers relying on specific models need not worry, as they can still call GPT-4, GPT-3.5, DALL-E 2, and other old models via API, ensuring development flexibility.
  • Historical Achievements: Since its release in March 2023, GPT-4 defined industry benchmarks with its near-human-level multimodal capabilities, profoundly impacting the AI landscape over the past two years.

Researcher’s Thoughts

  • For Practitioners: GPT-4’s approximately two-year service period in ChatGPT highlights the rapid iteration of AI models; practitioners need to stay sharp and quickly adapt to new technologies. OpenAI’s dual-track strategy for front-end (ChatGPT mandatory upgrade) and back-end (API retains old versions) is noteworthy, but also reminds developers dependent on specific APIs to prepare for long-term maintenance or eventual migration.
  • For Ordinary People: Users will naturally transition to the more capable GPT-4o, potentially getting a more accurate and smooth experience in daily interactions. This adjustment has little impact on free users, as they primarily use a simplified version of GPT-4o.

Recommended Reading

02

🚀 AI Agent Genspark Revenue Generation Myth: Achieved Ten Million Dollars ARR in 9 Days

The AI agent company MainFunc, co-founded by former Baidu VP Jing Kun and former Xiaodu CTO Zhu Kaihua, and its product Genspark have recently attracted industry attention. Founder Jing Kun announced that its core product “Super Agent,” within just 9 days of its official launch, achieved an Annual Recurring Revenue (ARR) as high as 10 million US dollars. Genspark Super Agent is a general-purpose AI agent capable of independently thinking, planning, and invoking tools to complete users’ daily tasks. Its features include fast interaction speed, low error rate, and allowing users to guide adjustments during execution. Even more attractive is its strategy of being free and not requiring an invitation code, greatly lowering the barrier to entry.

Core Highlights

  • Astonishing Speed: Genspark announced its AI agent product surpassed $10 million ARR just 9 days after launch, demonstrating strong early market appeal.
  • Free Strategy: The product emphasizes free use and no invitation code; this low-barrier strategy greatly promotes user growth and potential business conversion.
  • Powerful Functionality: Adopts an innovative “hybrid agent system,” integrating multiple LLMs and self-developed toolsets, capable of autonomously handling complex tasks like travel planning and email sending, emphasizing real-time capability and reliability.
  • Star Team: The founding team consists of former Baidu T11 level executives, possessing top-tier technical backgrounds and rich product experience.

Researcher’s Thoughts

  • For Practitioners: Genspark’s achievement (if the data is accurate and sustainable) is a strong boost for the AI agent track, indicating strong market demand and willingness to pay for agents that can automate complex tasks. It sets a high bar for other startups and intensifies competition. Its market entry strategy of “powerful product + extremely low barrier” is worth studying.
  • For Ordinary People: Provides an opportunity to try a powerful AI assistant for free, which can help handle various matters from planning to creation, offering an early experience of the potential of next-generation personal productivity tools. If its claimed low error rate holds true, it will greatly enhance the user experience.

Recommended Reading

03

🚀 xAI Officially Rows the Oar! Grok 3 API Lands, Directly Facing OpenAI Legal Battle Fire

Elon Musk’s artificial intelligence company xAI officially released the API for its flagship large model Grok 3 around April 10, 2025. This move not only marks xAI’s formal entry into the commercial AI API market, competing with giants like OpenAI and Google, but also occurs at a sensitive moment when the legal battle between Musk and OpenAI is escalating. This release provides Grok 3 and its lightweight version Grok 3 Mini, both touted as having powerful “reasoning” capabilities, and hints at the potential for future integration into Musk’s broader technology ecosystem (such as X, Tesla).

Core Highlights

  • Formal Release: xAI launched the API for the Grok 3 model, opening its flagship AI model to global developers and enterprises.
  • Dual Models Released Simultaneously: Simultaneously offers two versions: the high-performance Grok 3 and the more cost-effective Grok 3 Mini.
  • Legal Background: The API release timing coincides with a sharp escalation in the legal dispute between Musk and OpenAI, with OpenAI filing a countersuit against Musk.
  • Ecosystem Ambition: Grok already provides support for the X platform; its API-ization opens up possibilities for future integration into businesses under Musk like Tesla, SpaceX, etc.

Researcher’s Thoughts

  • For Practitioners/Developers: Offers a new large model API choice, particularly attractive for developers focusing on “reasoning” capabilities or hoping to leverage X platform’s real-time data. However, attention should be paid to its pricing and yet-to-be-verified performance stability. This also suggests Musk might be building a tighter cross-company technology ecosystem.
  • For Ordinary People: In the future, might see Grok-powered features in more applications. The competition and legal disputes between Musk and OpenAI also give ordinary users a more direct understanding of the direction of AI development, and the tension between openness and commercialization.

Recommended Reading

04

🛠️ Google Cloud Next Big Explosion: Agent Development Kit, AI Assistant, and New Model Chip Released Simultaneously

At the recent Google Cloud Next 2025 conference, Google showcased its comprehensive investment in AI, releasing a series of new products and updates covering developer tools, AI models, and underlying infrastructure. Core releases include the open-source Agent Development Kit (ADK), Gemini Code Assist updates with agent capabilities, Firebase Studio integrating Gemini for cloud development environments, the cost-effective new model Gemini 2.5 Flash, and the next-generation TPU chip “Ironwood” designed for inference, fully promoting the Agentic AI computing paradigm.

Core Highlights

  • Open Source Framework: Released and open-sourced the Agent Development Kit (ADK), a Python framework for building complex, multi-agent systems, simplifying the process from development to deployment.
  • Intelligent Coding: Gemini Code Assist (preview) gained “agent” capabilities, able to autonomously execute multi-step programming tasks, such as generating applications based on requirement documents, code conversion, automated testing, etc.
  • Cloud Development: Launched Firebase Studio (preview), a browser-based IDE based on VS Code and Gemini, supporting zero-setup start for full-stack AI application development.
  • Flash Model: Launched the Gemini 2.5 Flash model, complementing the Pro version, optimized for speed and cost-effectiveness, suitable for scenarios requiring fast responses and large-scale deployment.
  • Ironwood Chip: Released the seventh-generation TPU “Ironwood,” designed for the first time specifically for large-scale AI inference, significantly improving memory capacity, bandwidth, and energy efficiency ratio compared to the previous generation.

Researcher’s Thoughts

  • For Practitioners: Google is building a complete development ecosystem around “Agentic AI,” lowering the barrier through open-source ADK, integrated development environments, and intelligent coding tools to attract developers to its cloud platform. Gemini 2.5 Flash provides choices meeting different cost and performance needs. The self-developed Ironwood TPU strengthens Google’s differentiated competitive advantage in hardware infrastructure.
  • For Ordinary People: Although primarily aimed at developers and enterprises, it may ultimately bring indirect benefits: more powerful development tools accelerate AI application innovation; more efficient models and chips could make future AI services more powerful, faster responding, and potentially even lower cost.

Recommended Reading

Today’s Summary

  • Model Updates: OpenAI officially announced GPT-4 will be fully replaced by GPT-4o in ChatGPT next year, simultaneously hinting at potential new developments like GPT-4.1, million-level context, and enhanced memory.
  • Agent Commercialization: AI agent startup Genspark achieved ten million dollars ARR just 9 days after product launch, strongly validating the market demand and commercial value of high-performance agents.
  • Development Ecosystem Building: Google Cloud released an open-source agent development kit (ADK), Firebase Studio integrating AI, and intelligent code assistants, aiming to create an end-to-end development platform for AI agents.
  • New API Debut: Elon Musk’s xAI company released the Grok 3 API, challenging the AI market with high-end pricing and unique X platform real-time data integration.

About Us

🚀 Leading enterprise digital transformation, shaping the industry future together. We specialize in building customized digital systems integrating AI, achieving intelligent upgrades and deep integration of business processes. Relying on a core team with top tech backgrounds from MIT, Microsoft, etc., we help you build powerful AI-driven infrastructure, enhance efficiency, drive innovation, and achieve industry leadership status.