OpenAI Launches GPT-5.3 Instant While Google Unveils Gemini 3.1 Flash-Lite

GPT-5.3 Instant: A More Natural ChatGPT

OpenAI has integrated its new GPT-5.3 Instant model into ChatGPT, focusing on refining the tone, relevance, and flow of conversations. According to the company, the update is designed to make everyday interactions with the chatbot more useful and natural.

The predecessor, GPT-5.2 Instant, was known for occasionally delivering responses that felt "overly cautious and preachy" on sensitive topics. The updated model significantly reduces such moralizing preambles.

When processing web-sourced information, GPT-5.3 Instant produces more coherent and well-structured answers. The model is better at reading between the lines of user queries and cross-referencing retrieved data with its own knowledge base and reasoning capabilities.

Key improvements include:

• More natural conversational style — unnecessary phrases like "Stop. Take a breath" have been eliminated;
• Reduced hallucination frequency;
• Better text generation quality.

OpenAI cautioned that responses in some languages may still sound overly literal. GPT-5.3 Instant is already available to all ChatGPT users and developers via API. Support for GPT-5.2 Instant will continue until June 3, 2026.

OpenAI Building a GitHub Competitor

In parallel, multiple media outlets report that OpenAI is developing its own code hosting platform — essentially a GitHub alternative. The project is still in its early stages, but the strategic decision has been made. The service is expected to operate on a paid subscription model, though no additional details have been disclosed.

Recurring GitHub outages that disrupted OpenAI's engineering workflows are reportedly among the motivations for building the product. Launching such a platform would put OpenAI in direct competition with its major shareholder Microsoft, which owns GitHub.

In late February, OpenAI raised $110 billion in funding at a $730 billion valuation — one of the largest investment rounds in startup history.

Why This Matters

The simultaneous release of upgraded models from two AI market leaders underscores the intensifying competition for developers and enterprise clients. OpenAI is prioritizing conversation quality and ecosystem expansion (even building its own code hosting service), while Google is competing on price and speed. For businesses and developers, this translates to a wider array of tools and lower costs for accessing state-of-the-art neural networks.

Gemini 3.1 Flash-Lite: Google's Most Cost-Effective Model

Google has released a preview of Gemini 3.1 Flash-Lite, positioning it as the fastest and most affordable model in the Gemini 3 family. Pricing comes in at $0.25 per million input tokens and $1.50 per million output tokens.

Flash-Lite is optimized for building AI agents and handling tasks at scale: translating large data volumes, content moderation, and generating user interfaces. According to independent researchers at Artificial Analysis, the new model processes information 2.5 times faster than Gemini Flash 2.5.

The 3.1 Flash-Lite preview is available to developers through the Gemini API and Google AI Studio. Enterprise users can access it via Vertex AI.

Earlier in February, Google released Gemini 3.1 Pro — an updated model that set new benchmark records. Before that, the company upgraded its Gemini 3 Deep Think reasoning mode, designed for tackling complex scientific and engineering problems.