OpenAI upgrades GPT-5.5 Instant with improved context tracking and location awareness

OpenAI has released an update to GPT-5.5 Instant, the default model powering ChatGPT since May, with improvements to context understanding and multi-turn conversation handling.

What changed

The updated model now better identifies the underlying goal of queries and maintains context across multiple conversation turns. When users add conditions, clarifications, or push back on responses, GPT-5.5 Instant adapts more effectively rather than repeating previous answers.

The model also now handles complex multi-part questions more reliably, addressing multiple points in a single query instead of ignoring some components.

New location features

GPT-5.5 Instant now includes location context understanding, allowing it to provide geographically relevant recommendations. Users asking for restaurant suggestions while traveling should receive options actually nearby, rather than generic recommendations. The feature extends to product recommendations and business information.

The update also enables the chatbot to surface images when relevant to the query, though OpenAI did not specify how this image selection works.

Previous benchmark performance

When OpenAI made GPT-5.5 Instant the ChatGPT default in May, the company reported the model produced 52.5 percent fewer hallucinated statements and 37.3 percent fewer factual errors during internal testing compared to its predecessor. No new benchmark scores were provided with this update.

Availability

The improvements are live now for all ChatGPT users, as GPT-5.5 Instant remains the default model across free and paid tiers.

What this means

This update focuses on conversational quality rather than raw capability increases. The context tracking improvements address a persistent chatbot limitation: models losing thread of what users actually want as conversations evolve. The location awareness feature suggests OpenAI is positioning ChatGPT more directly against traditional search engines and recommendation services like Google Maps and Yelp, though the effectiveness of these features remains to be verified through independent testing.