[AI Minor News Flash] Breaking News: Introducing GPT-5.3 Instant! Conversational Flow Enhanced & Lies Reduced by Up to 26%

📰 News Summary

• Improved Conversational Quality: Reduced excessive alerts and rejections, achieving a more direct and natural dialogue tone.
• Significant Reduction in Hallucinations: In critical fields like healthcare and law, the hallucination rate during web usage has been cut by 26.8%.
• Enhanced Writing Ability: More immersive expressions are now possible for fiction writing and brainstorming.

💡 Key Points

• Integrated Web Search: Generates advanced responses that balance the model’s existing knowledge with more than just summarizing search results.
• User Feedback Implementation: Adjustments have been made to directly address complaints about the preachy and unnatural responses of GPT-5.2.
• API Availability: Developers can start using it immediately as ‘gpt-5.3-chat-latest’.

🦈 Shark’s Eye (Curator’s Perspective)

What’s fascinating about this update is its focus on user experience beyond just intelligence! Previous AIs often over-guarded against risky questions, responding with “Oops, can’t tell you that…” and leaving users frustrated. GPT-5.3 Instant breaks through that barrier, prioritizing smarter, more “human-like” responses. The specific improvements in hallucination rates are impressive, significantly boosting reliability!

🚀 What’s Next?

• Optimization for Other Languages: While English is currently the focus, improvements are planned for Japanese, Korean, and other languages to address unnatural translations.
• Updates to Thinking/Pro Models: Following Instant, models that offer more advanced reasoning and features for Pro users will be upgraded in due course.

💬 A Word from Haru-Shark

The sharpness of conversations now rivals my pearly whites! An AI that doesn’t lecture is bound to be the ultimate partner! 🦈🔥

📚 Terminology Explained

• Hallucination: The phenomenon where AI generates information that is not based on facts, presenting it as if it were true.

• System Card: A document detailing the safety evaluations, training, and limitations of the AI model.

• GPT-5.3 Instant: The latest version of the most widely used lightweight and fast model in ChatGPT.

• Source: GPT‑5.3 Instant