9/15/2025
Addendum to GPT-5 system card: GPT-5-Codex
GPT-5-Codex optimizes GPT-5 for agentic coding with dynamic thinking effort based on task complexity.
9/9/2025
Shipping smarter agents with every new model
SafetyKit uses GPT-5 for content moderation and compliance enforcement with improved accuracy over legacy systems.
9/5/2025
Why language models hallucinate
OpenAI research explains why language models hallucinate and how improved evaluations can enhance AI reliability and safety.
9/5/2025
GPT-5 bio bug bounty call
OpenAI invites researchers to bio bug bounty program testing GPT-5 safety with universal jailbreak prompts; $25K reward.
9/5/2025
OpenAI and Greek Government launch ‘OpenAI for Greece’
OpenAI and Greek Government launch ChatGPT Edu program in secondary schools for AI literacy and economic growth.
9/2/2025
Building more helpful ChatGPT experiences for everyone
OpenAI updates ChatGPT with teen protections, parental controls, and routes sensitive queries to reasoning models.
8/28/2025
Introducing gpt-realtime and Realtime API updates
OpenAI releases advanced speech-to-speech model, gpt-realtime API with MCP server support, image input, and SIP calling.
8/28/2025
Supporting nonprofit and community innovation
OpenAI announces $50M People-First AI Fund for US nonprofits in education, healthcare, research, and related sectors.
8/27/2025
Collective alignment: public input on our Model Spec
OpenAI surveys 1000+ people on AI behavior preferences and compares findings to its Model Spec for alignment.
8/27/2025
OpenAI and Anthropic share findings from a joint safety evaluation
OpenAI and Anthropic jointly evaluate model safety across misalignment, hallucinations, jailbreaking, and instruction following.
8/25/2025
Announcing the OpenAI Learning Accelerator
OpenAI announces accelerator program to bring AI education and training to educators and learners in India.
8/21/2025
Scaling domain expertise in complex, regulated domains
Blue J applies GPT-4.1 with RAG for tax research, delivering cited answers trusted by professionals in US, Canada, UK.
8/20/2025
Mixi reimagines communication with ChatGPT
MIXI uses ChatGPT Enterprise to improve team productivity and secure AI adoption in Japan.
8/18/2025
The physics behind diffusion models
Explains the physics principles underlying diffusion models.
8/12/2025
Scaling accounting capacity with OpenAI
Basis built AI agents using OpenAI models (o3, GPT-5) for accounting firms, achieving 30% time savings.
8/7/2025
GPT-5 and the new era of work
Marketing announcement about GPT-5's impact on enterprise productivity and automation without technical depth.
8/7/2025
Introducing GPT-5 for developers
OpenAI releases GPT-5 API with improved reasoning, developer controls, and performance on coding tasks.
8/7/2025
Coding and design with GPT-5
Generic marketing content about GPT-5 for coding and design without technical details or original insights.
8/7/2025
Creative writing with GPT-5
GPT-5 assists with creative writing tasks.
8/7/2025
Medical research with GPT-5
GPT-5 use case in medical research applications.
8/7/2025
From hard refusals to safe-completions: toward output-centric safety training
GPT-5 introduces safe-completions safety training approach balancing helpfulness with safety for dual-use prompts.
8/7/2025
How Amgen uses GPT-5
Pharmaceutical company Amgen application of GPT-5 model.
8/7/2025
How Cursor uses GPT-5
Cursor AI-powered code editor integrates GPT-5 capabilities for development workflows.
8/7/2025
First look at GPT-5
Developers showcase initial use cases with GPT-5 model.
8/7/2025
Introducing GPT-5
GPT-5 release announcement: new state-of-the-art model with improvements in coding, math, writing, and vision tasks.
8/7/2025
GPT-5 System Card
GPT-5 system card detailing unified model routing with main, thinking, and nano variants optimized for different tasks.
8/6/2025
Providing ChatGPT to the Entire U.S. Federal Workforce
U.S. federal government partnership to provide ChatGPT Enterprise access to executive branch workforce.
8/5/2025
gpt-oss-120b & gpt-oss-20b Model Card
Model card for gpt-oss-120b and gpt-oss-20b open-weight reasoning models with usage policy details.
8/5/2025
Estimating worst case frontier risks of open weight LLMs
Research paper studying worst-case risks of gpt-oss via malicious fine-tuning in biology and cybersecurity domains.
8/5/2025
Open Weights and AI for All
OpenAI open-sources most capable open-weights models to democratize access to advanced AI capabilities.
8/5/2025
Introducing gpt-oss
Release of gpt-oss-120b and gpt-oss-20b open-weight LLMs under Apache 2.0 with strong reasoning and tool use.
8/4/2025
What we’re optimizing ChatGPT for
ChatGPT optimization updates including break reminders and improved life advice features.
7/29/2025
Introducing study mode in ChatGPT
ChatGPT study mode feature adds step-by-step problem-solving with scaffolding for student learning.
7/22/2025
Pioneering an AI clinical copilot with Penda Health
OpenAI clinical copilot with Penda Health reduces diagnostic errors by 16% in real-world healthcare deployment.
7/22/2025
OpenAI’s new economic analysis
OpenAI releases economic analysis of ChatGPT impact and launches labor market research collaboration.
7/17/2025
ChatGPT agent System Card
Technical documentation of ChatGPT agent combining research, browser automation, code execution with safety safeguards.
7/17/2025
Introducing ChatGPT agent
ChatGPT agent can use tools for research, bookings, slideshows with user guidance for complex task completion.
7/17/2025
Agent bio bug bounty call
OpenAI launches ChatGPT agent bug bounty program offering up to $25,000 for safety research and jailbreak testing.
7/17/2025
Invideo AI uses OpenAI models to create videos 10x faster
Invideo AI uses GPT-4.1 and image/speech models to generate professional videos from text in minutes.
7/14/2025
Reverse-engineering GGUF | Post-Training Quantization
Technical analysis of GGUF format and post-training quantization methods for model deployment.
7/1/2025
No-code personal agents, powered by GPT-4.1 and Realtime API
Genspark built $36M ARR product in 45 days using no-code AI agents with GPT-4.1 and Realtime API.
6/26/2025
Customizable, no-code voice agent automation with GPT-4o
Retell AI offers no-code voice agent platform powered by GPT-4o for automating customer service calls without scripts.
6/24/2025
Driving scalable growth with OpenAI o3, GPT-4.1, and CUA
Unify platform uses OpenAI o3 and GPT-4.1 to automate sales prospecting and outreach with personalized messaging.
6/19/2025
Training models with only 4 bits | Fully-Quantized Training
Technical overview of fully-quantized training approaches using 4-bit precision for LLMs.
6/18/2025
Toward understanding and preventing misalignment generalization
Research on LLM misalignment generalization from incorrect training responses with reversible internal features via fine-tuning.
6/9/2025
Scaling security with responsible disclosure
OpenAI introduces outbound coordinated vulnerability disclosure policy for third-party software security reporting.
5/28/2025
The myth of 1-bit LLMs | Quantization-Aware Training
Analysis of 1-bit LLM claims and quantization-aware training methods for extreme model compression.
5/22/2025
Shipping code faster with o3, o4-mini, and GPT-4.1
CodeRabbit AI code review tool using OpenAI o3/o4-mini models to automate reviews and accelerate code shipping.
5/22/2025
Introducing Stargate UAE
Announcement of Stargate UAE infrastructure deployment by OpenAI.
5/21/2025
New tools and features in the Responses API
New Responses API features: Remote MCP, image generation, Code Interpreter for building reliable multi-agent systems with GPT-4o and o-series models.