OpenAI has released GPT‑5.1, an update to its GPT‑5 series, introducing improvements in reasoning, conversation quality, and user personalization. The update is named GPT‑5.1 to reflect iterative improvements while remaining within the GPT‑5 generation.
The release includes two models, GPT‑5.1 Instant and GPT‑5.1 Thinking, representing a step toward a ChatGPT experience that is more adaptable, easier to use, and continuously improving in both intelligence and usability.
GPT‑5.1 Instant
GPT‑5.1 Instant, the most commonly used ChatGPT model, now responds in a warmer and more conversational tone. Instruction following has been improved, allowing the model to answer user queries more accurately.
Adaptive reasoning enables GPT‑5.1 Instant to allocate time for more challenging questions, resulting in more detailed and precise responses while maintaining fast reply times. Early tests show performance gains in math and coding evaluations, including AIME 2025 and Codeforces.
GPT‑5.1 Thinking
GPT‑5.1 Thinking, designed for complex reasoning, adapts its processing time to the complexity of questions. It delivers faster responses for simple queries and more thorough answers for complex tasks.
Compared with GPT‑5 Thinking, it is roughly twice as fast on easier tasks and twice as slow on the most difficult ones. Responses are clearer, with reduced technical jargon, and the default tone is warmer and more empathetic.
Key Features of GPT‑5.1
- Adaptive Model Routing: Automatically directs queries to the model best suited for the task.
- Six Chat Styles: Provides Default, Friendly, Efficient, Professional, Candid, and Quirky options.
- Legacy Styles Retained: Cynical and Nerdy options remain available.
- Fine-Tuning Controls: Allows adjustments for conciseness, warmth, scannability, and emoji usage.
- Proactive Suggestions: ChatGPT can offer to update tone and style during conversations.
- Instant Effect: Personalization updates apply immediately across all chats, including ongoing conversations.
- Improved Custom Instruction Adherence: Provides more precise control over response tone and behavior, applicable to both Instant and Thinking models.
Looking Ahead
OpenAI plans to continue enhancing GPT‑5.1 in both intelligence and usability. Future updates are expected to further improve reasoning, conversational quality, and personalization features, ensuring that ChatGPT becomes increasingly adaptable to individual user preferences.
Availability
GPT‑5.1 Instant and GPT‑5.1 Thinking are rolling out starting with paid users (Pro, Plus, Go, Business), followed by free and logged-out users.
- Enterprise and Education plans have a seven-day early-access toggle (off by default).
- GPT‑5.1 will later become the default model for all users.
- Legacy GPT‑5 models remain available to paid users for three months.
- GPT‑5.1 Instant and GPT‑5.1 Thinking will be added to the API as
gpt-5.1-chat-latestandgpt-5.1, both supporting adaptive reasoning.