gpt-5.1-2025-11-13

gpt-5.1 Model Version as of November 13, 2025

Basic Information

GPT-5.1 is the next-generation model in the GPT-5 series released by OpenAI on November 13, 2025, via its API platform. It is designed for developers to adapt to agent and coding tasks and has been made available to all paid API tiers, with pricing and rate limits consistent with GPT-5. Simultaneously launched are gpt-5.1-codex and gpt-5.1-codex-mini, the former optimized for long-term coding tasks. The model supports both Responses and Chat Completions APIs, offering a “no-reasoning” mode and advanced prompt caching capabilities. It prioritizes delivering faster performance than GPT-5 for customers, while the previous GPT-5 version will be retained to ensure a smooth transition.

Core Features

It features adaptive reasoning capabilities, dynamically adjusting thinking time based on task complexity—speeding up simple tasks while saving tokens, and maintaining high reliability for complex tasks. A new apply_patch tool has been introduced, enabling reliable code editing through structured diffs without requiring JSON escaping. It also supports shell tools, allowing users to run Shell commands and implement a “plan—execute” loop. Its coding capabilities have been upgraded, enhancing controllability and code quality; it achieves 76.3% on SWE-bench Verified tests and adapts well to diverse coding scenarios.

Technical Highlights

The “no-reasoning” mode has been introduced, setting reasoning_effort='none'. This mode delivers fast responses in low-latency scenarios and supports high-performance tool calls. Advanced prompt caching can be retained for up to 24 hours, reducing input token costs by 90% and improving multi-turn interaction efficiency. Token usage has been optimized—token consumption for tool-intensive tasks is only half that of competitors, while maintaining high accuracy.

Market Impact

It helps developers build efficient agent workflows, boosting efficiency in scenarios such as coding and tool calls. It reduces AI application development costs by optimizing caching and token usage, thereby lowering expenses. It accelerates the deployment of low-latency, high-intelligence AI applications, adapting to needs across multiple industries. Its tools and reasoning capabilities set a benchmark for model efficiency and practicality across industries.

Comparison with GPT-5:

1. GPT-5.1’s default settings are more friendly and natural, with stronger conversational abilities;
2. Its ability to follow instructions has also become significantly stronger;
3. It is the first to employ adaptive reasoning technology. Before answering more challenging questions, it can autonomously decide when to engage in thinking, thus providing more comprehensive and accurate answers while maintaining rapid response times.