OpenAI Releases Model With Less Hallucinations And Improved Accuracy

OpenAI has rolled out GPT-5.3 Instant, a new upgrade to ChatGPT designed to tackle one of the biggest complaints about artificial intelligence: inaccurate or awkward answers.

The update focuses on making the chatbot more reliable, with the company promising responses that are “more accurate and less embarrassing.”

The move comes as the global AI race intensifies, where credibility and response quality are becoming just as important as raw technological power.

Rather than introducing flashy new capabilities, GPT-5.3 Instant focuses on precision. According to OpenAI’s internal benchmarks, the update reduces factual errors by 26.8% compared with GPT-5.2 Instant, marking a significant improvement in the model’s ability to deliver reliable answers.

The company has also refined how the model responds to user queries. ChatGPT is now less likely to produce overly cautious refusals when responding to legitimate prompts, while improvements to language calibration help generate smoother and more natural phrasing.

At the same time, safety mechanisms remain in place, with the system better distinguishing between requests that genuinely require refusal and those that can be answered directly.

The result is an AI model designed to feel less rigid and less prone to the awkward or overly defensive responses that have sometimes frustrated users.

OpenAI confirmed that GPT-5.3 Instant is available immediately to all ChatGPT users, including both free and paid accounts. Developers can also access the upgraded model through OpenAI’s API under the name “gpt-5.3-chat-latest.”

The previous version, GPT-5.2 Instant, will remain accessible for paying users in the Legacy Models section for a limited time before being retired on June 3, 2026.

OpenAI Focuses On Reliability In The AI Race

The release highlights a broader strategy from OpenAI: improving trust in AI systems rather than simply unveiling dramatic new features.

With ChatGPT continuing to dominate the generative AI landscape and reportedly attracting the majority of user traffic among chatbot platforms, even incremental improvements to accuracy can have a major impact.

For developers and businesses integrating AI tools into professional workflows, fewer hallucinations and more consistent responses could significantly reduce the need for complex prompt engineering and manual verification.

Instead of chasing spectacle, GPT-5.3 Instant signals a shift toward stability and credibility. In a rapidly evolving AI market, OpenAI appears to be betting that reliable tools — not just revolutionary ones — will determine who ultimately leads the industry.