OpenAI Debuts GPT-5.3, Fast-Tracks GPT-5.4 Amid User Feedback
Just under two hours after Google DeepMind unveiled Gemini 3.1 Flash-Lite, OpenAI swiftly launched a direct counterpunch. Late Tuesday local time, OpenAI rolled out its new model, GPT-5.3 Instant, in an overnight release. Moving beyond the typical benchmark-chasing updates, this iteration directly tackles one of ChatGPT users' most persistent frustrations: the overly cautious, "preachy" tone and cumbersome disclaimers.

For a long time, ChatGPT would often respond to certain prompts with a cold disclaimer or even a condescending refusal. GPT-5.3 Instant puts an end to that awkward dynamic. In official demos, when faced with a complex physics calculation request, the new model skips the "safety review" questions and instead provides parameters and formulas directly, adopting a more natural and collaborative tone. OpenAI acknowledged that a key focus for this version was fixing the problem of AI stalling conversations.
Beyond a major boost in conversational intelligence, GPT-5.3 Instant also delivers a substantial leap in practical performance. Data indicates the model's hallucination rate dropped by 26.8% in live online tests. Even when answering solely from its internal knowledge base, hallucinations decreased by nearly 20%. This significantly boosts AI's reliability in rigorous fields like medicine, law, and finance. Furthermore, its web search function now behaves more like a human researcher, synthesizing current news and background context into comprehensive analyses instead of just listing links.
Interestingly, GPT-5.3 Instant demonstrates a refined sense of "aesthetics" in creative writing. Where previous versions could feel mechanically emotional, the new model excels at crafting moving narratives through nuanced detail. GPT-5.3 Instant is now available on the ChatGPT website and app, with API access for developers. OpenAI also announced that the more powerful GPT-5.4 is already in development.
Related article
Xiaohongshu Restructures: Conan Named President, Creates AI Primary Department Dots and Overseas Division Rednote
On April 30, Xiaohongshu sent an internal memo to all employees announcing the launch of a new organizational restructuring. The core of this change involves fully integrating three business lines—community, e-commerce, and commercialization—along wi
Tencent's Xiaolongxia Surges Beyond Expectations, Team Expands Capacity 10x, Apologizes and Compensates
Tencent has officially launched WorkBuddy, an all-scenario AI intelligent agent, marking a new phase in the large model application layer race with high integration and a low deployment threshold.The product drew immediate industry attention on its l
Suno Lead Investor: Deleting Posts Won't Plug Copyright Lawsuit Hole
The much-anticipated AI music generation platform Suno is facing a tough copyright battle, and a candid remark from its lead investor may have handed the opposing side exactly the evidence they were hoping for. C.C. Gong, a partner at Menlo Ventures
Related Special Topic Recommendations
Comments (0)
0/500
Just under two hours after Google DeepMind unveiled Gemini 3.1 Flash-Lite, OpenAI swiftly launched a direct counterpunch. Late Tuesday local time, OpenAI rolled out its new model, GPT-5.3 Instant, in an overnight release. Moving beyond the typical benchmark-chasing updates, this iteration directly tackles one of ChatGPT users' most persistent frustrations: the overly cautious, "preachy" tone and cumbersome disclaimers.

For a long time, ChatGPT would often respond to certain prompts with a cold disclaimer or even a condescending refusal. GPT-5.3 Instant puts an end to that awkward dynamic. In official demos, when faced with a complex physics calculation request, the new model skips the "safety review" questions and instead provides parameters and formulas directly, adopting a more natural and collaborative tone. OpenAI acknowledged that a key focus for this version was fixing the problem of AI stalling conversations.
Beyond a major boost in conversational intelligence, GPT-5.3 Instant also delivers a substantial leap in practical performance. Data indicates the model's hallucination rate dropped by 26.8% in live online tests. Even when answering solely from its internal knowledge base, hallucinations decreased by nearly 20%. This significantly boosts AI's reliability in rigorous fields like medicine, law, and finance. Furthermore, its web search function now behaves more like a human researcher, synthesizing current news and background context into comprehensive analyses instead of just listing links.
Interestingly, GPT-5.3 Instant demonstrates a refined sense of "aesthetics" in creative writing. Where previous versions could feel mechanically emotional, the new model excels at crafting moving narratives through nuanced detail. GPT-5.3 Instant is now available on the ChatGPT website and app, with API access for developers. OpenAI also announced that the more powerful GPT-5.4 is already in development.
Xiaohongshu Restructures: Conan Named President, Creates AI Primary Department Dots and Overseas Division Rednote
On April 30, Xiaohongshu sent an internal memo to all employees announcing the launch of a new organizational restructuring. The core of this change involves fully integrating three business lines—community, e-commerce, and commercialization—along wi
Tencent's Xiaolongxia Surges Beyond Expectations, Team Expands Capacity 10x, Apologizes and Compensates
Tencent has officially launched WorkBuddy, an all-scenario AI intelligent agent, marking a new phase in the large model application layer race with high integration and a low deployment threshold.The product drew immediate industry attention on its l
Suno Lead Investor: Deleting Posts Won't Plug Copyright Lawsuit Hole
The much-anticipated AI music generation platform Suno is facing a tough copyright battle, and a candid remark from its lead investor may have handed the opposing side exactly the evidence they were hoping for. C.C. Gong, a partner at Menlo Ventures





Home






