はてなブックマーク

GPT-5.1, our newest flagship model, is designed to balance intelligence and speed for a variety of agentic and coding tasks, while also introducing a new none reasoning mode for low-latency interactions. Building on the strengths of GPT-5, GPT-5.1 is better calibrated to prompt difficulty, consuming far fewer tokens on easy inputs and more efficiently handling challenging ones. Along with these be

Oct 7, 2025Using PLANS.md for multi-hour problem solving Codex and the gpt-5-codex model can be used to implement complex tasks that take significant time to research, design, and implement. The approach described here is one way to prompt the model to implement these tasks and to steer it towards successful completion of a project. These plans are thorough design documents, and "living documents"

Think of prompting like briefing a cinematographer who has never seen your storyboard. If you leave out details, they’ll improvise – and you may not get what you envisioned. By being specific about what the “shot” should achieve, you give the model more control and consistency to work with. But leaving some details open can be just as powerful. Giving the model more creative freedom can lead to su

Important details about GPT-5-Codex and this guide: This model is not a drop-in replacement for GPT-5, as it requires significantly different prompting. This model is only supported with the Responses API and does not support the verbosity parameter. This guide is meant for API users of GPT-5-Codex and creating developer prompts, not for Codex users, if you are a Codex user refer to this prompting

Sep 9, 2025Context Engineering - Short-Term Memory Management with Sessions from OpenAI Agents SDK AI agents often operate in long-running, multi-turn interactions, where keeping the right balance of context is critical. If too much is carried forward, the model risks distraction, inefficiency, or outright failure. If too little is preserved, the agent loses coherence. Here, context refers to the

The OpenAI gpt-oss models are introducing a lot of new concepts to the open-model ecosystem and getting them to perform as expected might take some time. This guide is meant to help developers building inference solutions to verify their implementations or for developers who want to test any provider’s implementation on their own to gain confidence. Why is implementing gpt-oss models different? Th

Anoop Kotha(OpenAI), Julian Lee(OpenAI), Eric Zakariasson, et al. GPT-5, our newest flagship model, represents a substantial leap forward in agentic task performance, coding, raw intelligence, and steerability. While we trust it will perform excellently “out of the box” across a wide range of domains, in this guide we’ll cover prompting tips to maximize the quality of model outputs, derived from o

Aug 5, 2025Fine-tuning with gpt-oss and Hugging Face Transformers Authored by: Edward Beeching, Quentin Gallouédec, and Lewis Tunstall Large reasoning models like OpenAI o3 generate a chain-of-thought to improve the accuracy and quality of their responses. However, most of these models reason in English, even when a question is asked in another language. In this notebook, we show how OpenAI's open

Open-weight models are freely available base models that you can fine-tune or run locally.

The gpt-oss models were trained on the harmony response format for defining conversation structures, generating reasoning output and structuring function calls. If you are not using gpt-oss directly but through an API or a provider like Ollama, you will not have to be concerned about this as your inference solution will handle the formatting. If you are building your own inference solution, this g

Want to get OpenAI gpt-oss running on your own hardware? This guide will walk you through how to use Ollama to set up gpt-oss-20b or gpt-oss-120b locally, to chat with it offline, use it through an API, and even connect it to the Agents SDK. Note that this guide is meant for consumer hardware, like running a model on a PC or Mac. For server applications with dedicated GPUs like NVIDIA’s H100s, che

This cookbook serves as your practical guide to selecting, prompting, and deploying the right OpenAI model (between GPT 4.1, o3, and o4-mini) for specific workloads. Instead of exhaustive documentation, we provide actionable decision frameworks and real-world examples that help Solutions Engineers, Technical Account Managers, Partner Architects, and semi-technical practitioners quickly build worki

The GPT-4.1 family of models represents a significant step forward from GPT-4o in capabilities across coding, instruction following, and long context. In this prompting guide, we collate a series of important prompting tips derived from extensive internal testing to help developers fully leverage the improved abilities of this new model family. Many typical best practices still apply to GPT-4.1, s

Mar 11, 2025Doing RAG on PDFs using File Search in the Responses API Although RAG can be overwhelming, searching amongst PDF file shouldn't be complicated. One of the most adopted options as of now is parsing your PDF, defining your chunking strategies, uploading those chunks to a storage provider, running embeddings on those chunks of texts and storing those embeddings in a vector database. And t

Oct 14, 2024Custom LLM as a Judge to Detect Hallucinations with Braintrust Let's say you're working on a customer service bot and trying to evaluate the quality of its responses. Consider a question like "What is your return policy?" If the correct answer is "You can return items within 30 days of purchase," but your bot generates "You can return items within 30 days," how would you evaluate wheth

Oct 10, 2024Orchestrating Agents: Routines and Handoffs When working with language models, quite often all you need for solid performance is a good prompt and the right tools. However, when dealing with many unique flows, things may get hairy. This cookbook will walk through one way to tackle this. We'll introduce the notion of routines and handoffs, then walk through the implementation and show h

In this guide, we’ll explore how to use the o1 model, specifically o1-preview, to perform data validation through reasoning. We’ll walk through a practical example involving a synthetic medical dataset and demonstrate how to assess the model’s accuracy in identifying issues within the data. Overview Data validation is a critical step in ensuring the quality and reliability of datasets, especially

GPT-4o ("o" for "omni") and GPT-4o mini are natively multimodal models designed to handle a combination of text, audio, and video inputs, and can generate outputs in text, audio, and image formats. GPT-4o mini is the lightweight version of GPT-4o. Background Before GPT-4o, users could interact with ChatGPT using Voice Mode, which operated with three separate models. GPT-4o integrates these capabil

The new Assistants API is a stateful evolution of our Chat Completions API meant to simplify the creation of assistant-like experiences, and enable developer access to powerful tools like Code Interpreter and File Search. Chat Completions API vs Assistants API The primitives of the Chat Completions API are Messages, on which you perform a Completion with a Model (gpt-4o, gpt-4o-mini, etc). It is l

This notebook demonstrates how to use GPT's visual capabilities with a video. Although GPT-4.1-mini doesn't take videos as input directly, we can use vision and the 1M token context window to describe the static frames of a whole video at once. We'll walk through two examples: Using GPT-4.1-mini to get a description of a video Generating a voiceover for a video with GPT-4o TTS API from IPython.dis

GPT-5 Prompt Migration and Improvement Using the New Optimizer