OpenAI GPT-5.1 AMA and ChatGPT Market Performance

OpenAI revealed several functional advancements and safety plans during the GPT-5.1 AMA.
Key insights
Functional Advancements and Product Strategy of GPT-5.1
OpenAI stated in the AMA that the GPT-5.1 series includes multiple product improvements: GPT-5.1 Pro is set to be released soon, Thinking Mini will continue to be maintained, and the Canvas scenarios, writing, and coding adaptations are being optimized; at the API level, the Responses API allows caching to extend up to 24 hours and supports faster responses in reasoning effort=none mode with temperature/logprobs parameters. The model's knowledge cutoff remains September 2024, but it is labeled as June 2024 in the ChatGPT product, and the model typically avoids content beyond June in the product.
Development Progress of Multimodal, Voice, and Safety Mechanisms
OpenAI indicated that it is developing native multimodal capabilities (images can now be read/generated), but there is no timeline for deep audio and video input; the team emphasizes ongoing improvements to Voice Mode, but there is no confirmed release date for the GPT-5 series to enter Voice Mode. In terms of safety, the team is attempting to create more precise guardrails and localized rules based on user context, and plans to establish an age verification system to train the model to responsibly use age information, but there is no clear timeline for these developments.
Model Performance, Instruction Following, and Coding Capability Enhancements
OpenAI reported that GPT-5.1 performs overall better than GPT-5 in coding evaluations, achieving approximately 76% in SWE-bench Verified, improving quick responses to simple tasks and extending performance during deep reasoning; GPT-5.1 is more accurate in executing custom instructions, introducing a 'character personality' system and adjustable slider experiments; it has also optimized GPT-5.1-codex and the hybrid architecture of GPT-5.1-instant for long-term, agent-like coding scenarios.
Significant Decline in ChatGPT Market Share
Traffic statistics show that over the past 12 months, ChatGPT's market share has dropped from 86.6% to the current 72.3%, while Gemini has risen from 5.6% to 13.7%; products like DeepSeek, Grok, and Claude have also risen or rebounded at different times, indicating a trend of diversification in market share distribution.