Investing.com - OpenAI today released GPT-5.3 Instant, an upgraded version of the most commonly used ChatGPT model, designed to provide more accurate answers and improved conversational flow. The company stated that this update addresses user feedback regarding tone, relevance, and unnecessary refusals that affected the previous GPT-5.2 Instant model.
According to OpenAI’s internal evaluation, in high-risk areas such as medicine, law, and finance, the new model reduced hallucination rates by 26.8% when using web searches and by 19.7% when relying on internal knowledge. When users marked factual errors, hallucinations decreased by 22.5% when using the web and by 9.6% when not accessing the web.
GPT-5.3 Instant reduces unnecessary refusals and overly cautious responses. OpenAI said these issues previously caused earlier versions to appear “preachy” or “arrogant” on sensitive topics. The model now provides direct answers when appropriate and minimizes defensive openings and phrases like “stop, take a deep breath.”
This update improves web-based answers, better balancing online information with the model’s existing knowledge, rather than simply summarizing search results or providing long lists of links. OpenAI stated that the model can more effectively recognize underlying questions and prioritize important information.
GPT-5.3 Instant is now available to all ChatGPT users, with developers able to access it via API under the name “gpt-5.3-chat-latest.” GPT-5.2 Instant will continue to be available for three months for paid users’ legacy models until it is retired on June 3, 2026.
This article was translated with the assistance of artificial intelligence. For more information, please see our Terms of Use.
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
OpenAI upgrades ChatGPT and releases the GPT-5.3 Instant model to enhance accuracy
Investing.com - OpenAI today released GPT-5.3 Instant, an upgraded version of the most commonly used ChatGPT model, designed to provide more accurate answers and improved conversational flow. The company stated that this update addresses user feedback regarding tone, relevance, and unnecessary refusals that affected the previous GPT-5.2 Instant model.
According to OpenAI’s internal evaluation, in high-risk areas such as medicine, law, and finance, the new model reduced hallucination rates by 26.8% when using web searches and by 19.7% when relying on internal knowledge. When users marked factual errors, hallucinations decreased by 22.5% when using the web and by 9.6% when not accessing the web.
GPT-5.3 Instant reduces unnecessary refusals and overly cautious responses. OpenAI said these issues previously caused earlier versions to appear “preachy” or “arrogant” on sensitive topics. The model now provides direct answers when appropriate and minimizes defensive openings and phrases like “stop, take a deep breath.”
This update improves web-based answers, better balancing online information with the model’s existing knowledge, rather than simply summarizing search results or providing long lists of links. OpenAI stated that the model can more effectively recognize underlying questions and prioritize important information.
GPT-5.3 Instant is now available to all ChatGPT users, with developers able to access it via API under the name “gpt-5.3-chat-latest.” GPT-5.2 Instant will continue to be available for three months for paid users’ legacy models until it is retired on June 3, 2026.
This article was translated with the assistance of artificial intelligence. For more information, please see our Terms of Use.