OpenAI Blog

Standardizing OpenAI’s deep learning framework on PyTorch

We are standardizing OpenAI’s deep learning framework on PyTorch.

Scaling Kubernetes to 7,500 Nodes

We’ve scaled Kubernetes clusters to 7,500 nodes, producing a scalable infrastructure for large models like GPT-3, CLIP, and DALL·E, but also for rapid small-scale iterative research such as Scaling Laws for Neural Language Models.

Separating signal from noise in coding evaluations

A new analysis from OpenAI reveals issues in SWE-Bench Pro, a popular coding benchmark, raising concerns about reliability and accuracy in evaluating AI models.

Special projects

Impactful scientific work requires working on the right problems—problems which are not just interesting, but whose solutions matter.

OpenAI releases larger GPT-2 model

We’ve trained a large-scale unsupervised language model which generates coherent paragraphs of text, achieves state-of-the-art performance on many language modeling benchmarks, and performs rudimentary reading comprehension, machine translation, question answering, and summarization—all without task-specific training.

SimpleQA

A factuality benchmark called SimpleQA that measures the ability for language models to answer short, fact-seeking questions.

GPT-4o mini: advancing cost-efficient intelligence

OpenAI DayBreak – GPT-5.5-Cyber

OpenAI introduces new Daybreak tools, including Codex Security and GPT-5.5-Cyber, to help organizations find, validate, and patch vulnerabilities at scale.

Concrete AI Safety Problems

We (along with researchers from Berkeley and Stanford) are co-authors on today’s paper led by Google Brain researchers, Concrete Problems in AI Safety. The paper explores many research problems around ensuring that modern machine learning systems operate as intended.

New embedding models and API updates

OpenAI technical goals

OpenAI’s mission is to build safe AI, and ensure AI’s benefits are as widely and evenly distributed as possible.

Introducing Superalignment

We need scientific and technical breakthroughs to steer and control AI systems much smarter than us. To solve this problem within four years, we’re starting a new team, co-led by Ilya Sutskever and Jan Leike, and dedicating 20% of the compute we’ve secured to date to this effort. We’re looking for excellent ML…

Fine-tuning now available for GPT-4o

Fine-tune custom versions of GPT‑4o to increase performance and accuracy for your applications.

Research acceleration: The view inside OpenAI

Inside OpenAI, coding agents are reshaping AI research. Explore early data on agent usage, experiment velocity, task complexity, and research acceleration.

Responding to the next frontier of critical cyber capabilities

OpenAI is sharing preliminary cybersecurity evaluations for Astra and the steps we’re taking to strengthen safeguards and security controls.

Improvements to data analysis in ChatGPT

Improvements to data analysis in ChatGPT Interact with tables and charts and add files directly from Google Drive and Microsoft OneDrive.

Solving some formal math olympiad problems

We built a neural theorem prover for Lean that learned to solve a variety of challenging high-school olympiad problems, including problems from the AMC12 and AIME competitions, as well as two problems adapted from the IMO.

OpenAI Five: Goals and Progress

OpenAI Five is the first AI to beat the world champions in an esports game, having won two back-to-back games versus the world champion Dota 2 team, OG, at Finals this weekend. Both OpenAI Five and DeepMind’s AlphaStar had previously beaten good pros privately but lost their live pro matches, making this also the…

Developers can now submit apps to ChatGPT

Developers can now submit apps for review and publication in ChatGPT, with approved apps appearing in a new in-product directory for easy discovery. Updated tools, guidelines, and the Apps SDK help developers build powerful chat-native experiences that bring real-world actions into ChatGPT.

New Capabilities for GPT-3: Edit and Insert

We’ve released new versions of GPT-3 and Codex which can edit or insert content into existing text, rather than just completing existing text.

OpenAI Grove

Applications are now open for OpenAI Grove Cohort 2, a 5-week founder program designed for individuals at any stage, from pre-idea to product. Participants receive $50K in API credits, early access to AI tools, and hands-on mentorship from the OpenAI team.

Path to Astra: critical capabilities and frontier safeguards

Astra is the first OpenAI model to meet the Critical cybersecurity capability threshold under the Preparedness Framework, with stronger safeguards for release.

Vijaye Raji to become CTO of Applications with acquisition of Statsig

Vijaye Raji will step into a new role as CTO of Applications, reporting to CEO of Applications, Fidji Simo, following the acquisition of Statsig.

Introducing gpt-oss

We’re releasing gpt-oss-120b and gpt-oss-20b—two state-of-the-art open-weight language models that deliver strong real-world performance at low cost. Available under the flexible Apache 2.0 license, these models outperform similarly sized open models on reasoning tasks, demonstrate strong tool use capabilities, and…

HealthBench – An evaluation for AI systems and human health

HealthBench is a new evaluation benchmark for AI in healthcare which evaluates models in realistic scenarios. Built with input from 250+ physicians, it aims to provide a shared standard for model performance and safety in health.

Solving the Rubik’s cube with a robot hand

We’ve trained a pair of neural networks to solve the Rubik’s Cube with a human-like robot hand. The neural networks are trained entirely in simulation, using the same reinforcement learning code as OpenAI Five paired with a new technique called Automatic Domain Randomization (ADR). The system can handle situations it…

Pacing model development in an era of cyber-critical capabilities

OpenAI is strengthening monitoring, alignment, and security for frontier AI models. See how new safeguards are guiding the pace of model development.

Training agents to invent a language

In this post we’ll outline new OpenAI research in which agents develop their own language.

OpenAI and Microsoft

We’re working with Microsoft to start running most of our large-scale experiments on Azure.

Multimodal Neurons in Artificial Neural Networks

We’ve discovered neurons in CLIP that respond to the same concept whether presented literally, symbolically, or conceptually. This may explain CLIP’s accuracy in classifying surprising visual renditions of concepts, and is also an important step toward understanding the associations and biases that CLIP and similar…