OpenAI's new $20,000 per month pricing plan

Hello AI lovers! Here’s what you need to know about AI today:

👉 Google Search is adding a new AI mode feature

👉 OpenAI’s ridiculous $20K-per-month AI pricing plan

👉 AI agents are talking to each other in robo-language

and many more!

📧 Did someone forward you this email? Subscribe here for free to get the latest AI news everyday!

Read time: 5.1 minutes

What’s going on: Google has introduced an experimental "AI Mode" feature to its Search platform, aiming to compete with services like Perplexity AI and OpenAI’s ChatGPT Search. This new mode is designed to handle complex, multi-part questions and follow-ups directly within Google Search, offering users a more robust and interactive search experience. However, it is available initially to Google One AI Premium subscribers through Search Labs.

What does it mean: Unlike the existing AI Overviews, which provide concise summaries, AI Mode delivers full-page, AI-generated answers tailored to queries that might have previously required multiple searches. For instance, it can process intricate questions, perform detailed comparisons, and explore topics in depth, pulling information from various sources, including real-time data like the Knowledge Graph and shopping data.

More details: 

• AI Mode uses a custom version of Google’s Gemini 2.0 model, which brings advanced reasoning and multimodal capabilities to provide detailed, comprehensive responses.

• Google utilizes a “query fan-out” technique to gather and synthesize this information, ensuring responses are both thorough and verifiable, particularly in sensitive areas like health, where factuality is prioritized.

• Google is also upgrading AI Overviews with Gemini 2.0 for complex queries like coding and math.

• Interested in testing Google’s AI mode before anyone else? Join the waitlist.

What’s going on: OpenAI is planning to introduce a new range of “specialized” AI agents designed to handle complex tasks, with pricing that could reach as high as $20,000 per month. OpenAI is making a bold move by shifting from its previous highest subscription tier of $200 per month, which failed to generate profit, to a premium model targeting enterprise clients and high-income professionals. These AI agents will come in multiple tiers, including a $2,000-per-month option aimed at business professionals, a $10,000-per-month version for advanced software engineering tasks, and the top-tier $20,000-per-month agent boasting PhD-level research capabilities.

What does it mean: This development comes while OpenAI is receiving significant financial backing, with SoftBank committing $3 billion in 2025 and ongoing negotiations with Microsoft to convert its $14 billion investment into equity as OpenAI’s valuation increases to $157 billion. The company is betting that these advanced tools, capable of outperforming human teams in areas like sales lead sorting and intricate problem-solving, will justify their steep price tags and drive substantial long-term revenue, potentially accounting for 20% to 25% of OpenAI’s future income.

More details:

• Despite growing rapidly, OpenAI is still not profitable and its projected computing expenses could reach $37.5 billion annually by 2029.

• OpenAI has been consistently increasing its subscription prices, despite the existence of many other equally-good rivals such as DeepSeek-R1, Grok 3, Claude 3.7 Sonnet, and Gemini 2.0 Pro at relatively lower prices.

• OpenAI recently released its largest AI model, GPT-4.5, which is also very expensive to run, so expensive that the company says it’s evaluating whether to continue serving GPT-4.5 in its API in the long term. To access GPT-4.5’s API, OpenAI is charging developers $75 for every million input tokens (roughly 750,000 words) and $150 for every million output tokens.

• Interested in details? Read the original report on “The Information” website.

💬 OpenAI is rolling out its largest AI model yet, GPT-4.5, to ChatGPT Plus subscribers over the next 1-3 days, after initially launching it to ChatGPT Pro users last week.

🤖 GibberLink, a hackathon project created by Meta engineers, enables AI agents to recognize when they are speaking with each other and switch to a more efficient, non-human communication protocol called GGWave. Watch its viral video on X.

👨‍💻 A quarter of startups in Y Combinator's current cohort are using AI to generate 95% of their code. However, experts insist that more developers are needed to read, debug, and maintain the code.

🎙 Podcastle, a podcast recording and editing platform, has launched its own AI-powered text-to-speech model, Asyncflow v1.0, offering over 450 AI voices and an API for developers, aiming to compete with companies like ElevenLabs and Speechify.

🗣 Prime Video is testing AI-assisted dubbing in English and Latin American Spanish for 12 select movies and TV shows that previously lacked dubbing support, combining AI with human localization professionals to ensure quality control.

🍎 Apple is introducing AI-powered summaries of App Store reviews to provide users with a short paragraph highlighting key information from reviews in the iOS 18.4 update, currently in public beta.

👩‍🔬👨‍🔬 Experts are skeptical of recently-released Google's "AI co-scientist" and similar tools, arguing that they are overhyped and not particularly useful for guiding scientific discovery, citing issues with reproducibility, limited data.

💰 ARM Holdings has signed a $250 million deal with the Malaysian government over 10 years to support the country's chip design ecosystem, providing chip designs and technology while training 10,000 local engineers.

🐅 Google has open-sourced SpeciesNet, an AI model designed to identify animal species in camera trap images, trained on over 65 million images and capable of classifying them into more than 2,000 labels including animal species, taxa, and non-animal objects. Check out its GitHub repository.

AI + Checking Plagiarism

Elastic - Search - Principal Data Scientist

BILL - Senior Fraud Risk Data Scientist

Waymo - Senior Software Engineer, Applied ML and Evaluation

Oracle - AI/GPU Cloud Architect

Thank you for staying with us like always! If you are not subscribed, subscribe here for free to get more of these emails in your inbox! Cheers!