OpenAI Unveils GPT-5.3 Instant: A Step Forward Against Gemini and Claude

OpenAI Strengthens Its Offering with GPT-5.3 Instant

Amid tensions related to a controversial agreement with the Pentagon, OpenAI has launched GPT-5.3 Instant, a significant update to its flagship model. This model, which had already been integrated into Codex in February, is intended to gradually replace version GPT-5.2. Since the initial launch of GPT-5, which did not meet expectations, OpenAI has rolled out numerous small updates to retain its users. These users are increasingly being courted by competitors like Gemini, which is gaining popularity, and Claude, which stands out with its exclusive features. GPT-5.3 Instant is not merely a marketing operation but aims to address concrete issues faced by ChatGPT users.

Towards a More Flexible and Accurate Artificial Intelligence

The GPT-5.3 Instant model is now the default choice for text generation tasks that do not require complex reasoning. OpenAI also plans to launch a GPT-5.3 Thinking version soon. The Instant version is characterized by greater permissiveness compared to its predecessors. While GPT-5.2 avoided answering sensitive questions, particularly regarding health, GPT-5.3 Instant tackles these topics with a measured and non-judgmental approach.

As a research tool, GPT-5.3 Instant offers greater flexibility. Unlike GPT-5.2, which primarily relied on web sources, the new model prioritizes its internal knowledge before resorting to online data to support its responses. This approach is expected to create the impression of a more authentic intelligence, especially during complex analyses.

A Return to the Fundamentals of Generative AI

With GPT-5.3 Instant, OpenAI seems to be reconnecting with the principles that initially brought it success: a smarter and more human-like artificial intelligence. The tone adopted by the model is more natural and conversational, resembling what GPT-4o offered. Clichés and typical generative model automatons are diminished. Furthermore, the model ensures increased coherence from one conversation to another, giving users the impression of interacting with a consistent personality.

This shift towards more human-like foundations also translates into improved creative writing capabilities for ChatGPT. OpenAI teams have refined the model to produce texts with varied tones while maintaining narrative coherence. We can therefore expect more human-like writing, aiming to get closer to GPT-4.1. Additionally, the model continues to reduce its errors, particularly in medical, legal, and financial domains. OpenAI specifies that GPT-5.3 Instant reduces hallucinations by 26.8% when using the web and by 19.7% when relying solely on its internal knowledge.

Pricing and Mental Health Issues

Regarding the model's training, OpenAI remains discreet. The system card does not mention any significant methodological changes compared to GPT-5.2. However, the model operates in a sensitive area, particularly concerning the management of conversations about mental health. Without making it a selling point, OpenAI has introduced new evaluation criteria on topics such as self-harm, emotional dependency, and psychological distress, themes that have been the subject of much discussion recently.

In terms of pricing, OpenAI maintains the rates of GPT-5.2 Instant: $1.75 for input and $14.00 for output, per million tokens. The model is accessible via ChatGPT and the API under the identifier gpt-5.3-chat-latest. It is important to note that GPT-5.2-Instant will be disabled on June 3.

Although GPT-5.3 Instant does not revolutionize ChatGPT, in a race where Gemini and Claude are gaining popularity each week, even incremental improvements are crucial. For OpenAI, the goal is no longer to surprise but to maintain user satisfaction.