After ChatGPT, Grok 4 Announces Free Access, Launching a Free AI War?
On August 11, xAI announced on X platform that Grok 4 is now freely available to all users, with free users allowed limited daily queries, requiring subscription if exceeding limits. Paid users who pay 30 to 300 dollars monthly are crying out about a "subscription backstab".
Reporters from 'Science and Technology Board Daily' found through testing that free users can only use 3-5 Grok 4 queries every 12 hours, with only Grok 3 remaining available after reaching the limit. This restriction has sparked widespread user doubts, strongly pointing out the contrast between xAI's claimed "relaxed usage limits" and actual experience.
5 Free Queries in 12 Hours, Paid Users Cry "Backstabbed"
According to xAI, in automatic mode, Grok will automatically transfer complex queries to Grok 4; while "expert mode" allows continuous Grok 4 usage.
The reporter confirmed these function settings through hands-on testing. Additionally, when enabling "Quick Response" mode, it switches to Grok 3, while "Heavy" mode remains exclusively available to SuperGrok Heavy users.
[Image]
Reportedly, Grok offers two subscription plans: SuperGrok at 30 dollars/month and SuperGrok Heavy at 300 dollars/month. The free opening of previously paid Grok 4 has created a strong sense of "being backstabbed" among paid users. Users believe that Grok 4's widespread free access will weaken its subscription service value, questioning the reasonableness of 30 and 300 dollar monthly fees.
[Image]
One user directly questioned: "Why not upgrade SuperGrok members to limited version Grok 4 Heavy? Wouldn't I be wasting money on the upgrade?"
Regarding daily Grok 4 free usage quota, Grok's official X account previously replied that free users can use it up to 5 times every 12 hours, with users registered over 7 days and phone-verified getting 10 queries every 2 hours during Grok 4's temporary upgrade.
The reporter's test found that by the fourth query to Grok 4, the system refused to answer, prompting: "Your Grok usage limit is exhausted, please upgrade to SuperGrok." Only the "Quick Response" mode with Grok 3 remained functional, while "Automatic" and "Expert" modes were unavailable.
[Image]
Many users stated that xAI's "relaxed usage limits" description significantly differs from the actual provided usage times.
A user candidly told the reporter: "Understanding AI's high operating costs and accepting limits is fine. But the free times are insufficient. Sometimes before fully understanding a question, the 5 queries are exhausted, requiring a 12-hour wait. This mechanism is indeed unreasonable."
As of now, xAI and Musk have not provided an official response to users' "free access" doubts.
AI Free War Begins, Will You Choose ChatGPT-5 or Grok 4?
Grok 4's comprehensive free access comes just three days after GPT-5's release.
Facing OpenAI's flagship model opening to all free Team, Plus, Pro, and Team users, breaking the previous high-end AI model exclusivity to paid users, market voices suggest xAI's move might be a follow-up strategy to industry competitors.
Despite both playing the "free card", GPT-5 and Grok 4 have significant differences in technical paths and functional characteristics.
Specifically, OpenAI's ChatGPT-5 upgrade focuses on professional capability breakthroughs. In programming, it can efficiently generate visually coordinated website, app, and game prototypes from a single prompt. GPT-5 has also significantly reduced hallucination rates, especially in healthcare, providing personalized recommendations.
OpenAI CEO Sam Altman repeatedly emphasized GPT-5 can provide "dialogue experience with doctoral-level experts in various fields". Interestingly, when xAI released Grok 4 in July, Musk made a similar statement, claiming: "Grok 4 is better than doctoral level in all aspects".
Grok 4 demonstrates significant technical characteristics in deep reasoning abilities. According to authoritative HLE (Human Ultimate Exam) benchmark test data, its accuracy without tools is 25.4%, which can increase to 44.4% with multi-agent collaboration. The model excels in mathematics, achieving a perfect score in the American Mathematical Invitational Exam AIME 25. Additionally, its data processing architecture natively supports X platform real-time information flow integration, capable of automatically capturing social dynamics and event associations.
The free strategies implemented by OpenAI and xAI are reshaping the AI service value chain and innovation in profit models, pushing the user competition in the AI model industry to new heights.
As AI usage thresholds continue to lower, user expectations for functional depth and service experience will further rise. Amid the expanding emotional value demand wave, AI companion services are moving from marginal needs to the core battlefield of user competition.
The reporter noticed that facing users' growing emotional dependence on AI models, both giants have made strategic adjustments to attract more users and occupy market high ground.
OpenAI introduced four personality architectures in ChatGPT-5: Critic (sharp-tongued), Mechanical (efficient and direct), Listener (empathetic), and Tech Nerd (passionate explorer), allowing users to switch dialogue agents as needed. Additionally, responding to market demand for GPT-4o's anthropomorphic qualities, OpenAI quickly reintegrated the previously offline GPT-4o into the system.
On the other hand, xAI launched a companion AI represented by the two-dimensional virtual character Ani, transforming traditional tools into deep emotional companion carriers through high-fidelity personality modeling. Its core mechanism is building emotional attachment, ultimately achieving user stickiness through emotional value binding.
It's worth noting that OpenAI CEO Sam Altman revealed some concerns about users' excessive AI dependence.
He stated, "In the future, many people will truly trust ChatGPT's advice when making major decisions. While this prospect might be beautiful, it's also somewhat unsettling." Therefore, he believes it's necessary to actively consider strategies to ensure AI's overall benefits outweigh its drawbacks.
This article is from the WeChat public account "Science and Technology Board Daily", author: Li Jiayi, published by 36Kr with authorization.
