Research

<p><strong><a href="https://twitter.com/danveloper/status/2034353876753592372">Autoresearching Apple's "LLM in a Flash" to run Qwen 397B locally</a></strong></p> Here's a fascinating piece of research by Dan Woods, who managed to get a custom version of <a href="https://huggingface.co/Qwen/Qwen3.5-397B-A17B/tree/main">Qwen3.5-397B-A17B</a> running at 5.5+ tokens/second on a 48GB MacBook Pro M3 Max despite that model taking up 209GB (120GB quantized) on disk.</p> <p>Qwen3.5-397B-A17B is a Mixture

<p>OpenAI today: <a href="https://openai.com/index/introducing-gpt-5-4-mini-and-nano/">Introducing GPT‑5.4 mini and nano</a>. These models join GPT-5.4 which was released <a href="https://openai.com/index/introducing-gpt-5-4/">two weeks ago</a>.</p> <p>OpenAI's self-reported benchmarks show the new 5.4-nano out-performing their previous GPT-5 mini model when run at maximum reasoning effort. The new mini is also 2x faster than the previous mini.</p> <p>Here's how the pricing looks - all prices ar

GPT-5.4 mini and nano are smaller, faster versions of GPT-5.4 optimized for coding, tool use, multimodal reasoning, and high-volume API and sub-agent workloads.

<p><strong><a href="https://mistral.ai/news/mistral-small-4">Introducing Mistral Small 4</a></strong></p> Big new release from Mistral today (despite the name) - a new Apache 2 licensed 119B parameter (Mixture-of-Experts, 6B active) model which they describe like this:</p> <blockquote> <p>Mistral Small 4 is the first Mistral model to unify the capabilities of our flagship models, Magistral for reasoning, Pixtral for multimodal, and Devstral for agentic coding, into a single, versatile model.</p>

<p><strong><a href="https://claude.com/blog/1m-context-ga">1M context is now generally available for Opus 4.6 and Sonnet 4.6</a></strong></p> Here's what surprised me:</p> <blockquote> <p>Standard pricing now applies across the full 1M window for both models, with no long-context premium.</p> </blockquote> <p>OpenAI and Gemini both <a href="https://www.llm-prices.com/#sel=gemini-3-1-pro-preview-200k%2Cgpt-5.4-272k%2Cgemini-3-1-pro-preview%2Cgpt-5.4">charge more</a> for prompts where the token co

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/Workspace_Jan_Moment_Sheets_Blo.max-600x600.format-webp.webp">Today we announced new beta features for Gemini in Sheets to help you create, organize and edit entire sheets, from basic tasks to complex data analysis — just describe …

OpenAI introduces CoT-Control and finds reasoning models struggle to control their chains of thought, reinforcing monitorability as an AI safety safeguard.

Introducing GPT-5.4, OpenAI’s most most capable and efficient frontier model for professional work, with state-of-the-art coding, computer use, tool search, and 1M-token context.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/gemini-3.1_flash_Lite_blog_keyw.max-600x600.format-webp.webp">Gemini 3.1 Flash-Lite is our fastest and most cost-efficient Gemini 3 series model yet.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/BuildWith_Hero.max-600x600.format-webp.webp">Nano Banana 2 (Gemini 3.1 Flash Image) delivers Pro-level intelligence and fidelity for all image applications.

<img src="https://storage.googleapis.com/gweb-uniblog-publish-prod/images/0217_KeywordHeaderFinalc.max-600x600.format-webp.webp">Lyria 3 is now available in the Gemini app. Create custom, high-quality 30-second tracks from text and images.

A new preprint shows GPT-5.2 proposing a new formula for a gluon amplitude, later formally proved and verified by OpenAI and academic collaborators.

Introducing GPT-5.3-Codex-Spark—our first real-time coding model. 15x faster generation, 128k context, now in research preview for ChatGPT Pro users.

An autonomous lab combining OpenAI’s GPT-5 with Ginkgo Bioworks’ cloud automation cut cell-free protein synthesis costs by 40% through closed-loop experimentation.

GPT‑5.3-Codex is the most capable agentic coding model to date, combining the frontier coding performance of GPT‑5.2-Codex with the reasoning and professional knowledge capabilities of GPT‑5.2.

GPT-5.3-Codex is a Codex-native agent that pairs frontier coding performance with general reasoning to support long-horizon, real-world technical work.

By focusing on people, not pilots, the Bundesliga club is scaling efficiency, creativity, and knowledge—without losing its football identity.

Discover the Sora feed philosophy—built to spark creativity, foster connections, and keep experiences safe with personalized recommendations, parental controls, and strong guardrails.

On February 13, 2026, alongside the previously announced retirement⁠ of GPT‑5 (Instant, Thinking, and Pro), we will retire GPT‑4o, GPT‑4.1, GPT‑4.1 mini, and OpenAI o4-mini from ChatGPT. In the API, there are no changes at this time.

How Praktika uses GPT-4.1 and GPT-5.2 to build adaptive AI tutors that personalize lessons, track progress, and help learners achieve real-world language fluency

A data-driven report on how workers across industries use ChatGPT—covering adoption trends, top tasks, departmental patterns, and the future of AI at work.

Discover how Higgsfield gives creators cinematic, social-first video output from simple inputs using OpenAI GPT-4.1, GPT-5, and Sora 2.

Our latest report reveals stark differences in advanced AI adoption across countries and outlines new initiatives to help nations capture productivity gains from AI.

Tolan built a voice-first AI companion with GPT-5.1, combining low-latency responses, real-time context reconstruction, and memory-driven personalities for natural conversations.

GPT-5.2-Codex is OpenAI’s most advanced coding model, offering long-horizon reasoning, large-scale code transformations, and enhanced cybersecurity capabilities.

GPT-5.2-Codex is OpenAI’s most advanced coding model, offering long-horizon reasoning, large-scale code transformations, and enhanced cybersecurity capabilities.

OpenAI introduces a real-world evaluation framework to measure how AI can accelerate biological research in the wet lab. Using GPT-5 to optimize a molecular cloning protocol, the work explores both the promise and risks of AI-assisted experimentation.

OpenAI shipped Sora for Android in 28 days using Codex. AI-assisted planning, translation, and parallel coding workflows helped a nimble team deliver rapid, reliable development.

GPT-5.2 is OpenAI’s strongest model yet for math and science, setting new state-of-the-art results on benchmarks like GPQA Diamond and FrontierMath. This post shows how those gains translate into real research progress, including solving an open theoretical problem and generating reliable mathematical proofs.

GPT-5.2 is our most advanced frontier model for everyday professional work, with state-of-the-art reasoning, long-context understanding, coding, and vision. Use it in ChatGPT and the OpenAI API to power faster, more reliable agentic workflows.

GPT-5.2 is the latest model family in the GPT-5 series. The comprehensive safety mitigation approach for these models is largely the same as that described in the GPT-5 System Card and GPT-5.1 System Card. Like OpenAI’s other models, the GPT-5.2 models were trained on diverse datasets, including information that is publicly available on the internet, information that we partner with third parties to access, and information that our users or human trainers and researchers provide or generate.

UCLA Professor Ernest Ryu and GPT-5 solved a key question in optimization theory, showcasing AI’s role in accelerating mathematical discovery.

OpenAI introduces the first research cases showing how GPT-5 accelerates scientific progress across math, physics, biology, and computer science. Explore how AI and researchers collaborate to generate proofs, uncover new insights, and reshape the pace of discovery.

Introducing GPT-5.1-Codex-Max, a faster, more intelligent agentic coding model for Codex. The model is designed for long-running, project-scale work with enhanced reasoning and token efficiency.

OpenAI is exploring mechanistic interpretability to understand how neural networks reason. Our new sparse model approach could make AI systems more transparent and support safer, more reliable behavior.

GPT-5.1 is now available in the API, bringing faster adaptive reasoning, extended prompt caching, improved coding performance, and new apply_patch and shell tools.

We’re upgrading the GPT-5 series with warmer, more capable models and new ways to customize ChatGPT’s tone and style. GPT-5.1 starts rolling out today to paid users.

OpenAI introduces IndQA, a new benchmark for evaluating AI systems in Indian languages. Built with domain experts, IndQA tests cultural understanding and reasoning across 12 languages and 10 knowledge areas.

This system card details GPT-5’s improvements in handling sensitive conversations, including new benchmarks for emotional reliance, mental health, and jailbreak resistance.

Wrtn scaled AI apps to 6.5M users in Korea with GPT-5, creating ‘Lifestyle AI’ that blends productivity, creativity, and learning—now expanding across East Asia.

Our latest video generation model is more physically accurate, realistic, and controllable than prior systems. It also features synchronized dialogue and sound effects. Create with it in the new Sora app.

To address the novel safety challenges posed by a state-of-the-art video model as well as a new social creation platform, we’ve built Sora 2 and the Sora app with safety at the foundation. Our approach is anchored in concrete protections.

Sora 2 is our new state of the art video and audio generation model. Building on the foundation of Sora, this new model introduces capabilities that have been difficult for prior video models to achieve– such as more accurate physics, sharper realism, synchronized audio, enhanced steerability, and an expanded stylistic range.

Discover how SchoolAI, built on OpenAI’s GPT-4.1, image generation, and TTS, powers safe, teacher-guided AI tools for 1 million classrooms worldwide—boosting engagement, oversight, and personalized learning.

OpenAI invites researchers to its Bio Bug Bounty. Test GPT-5’s safety with a universal jailbreak prompt and win up to $25,000.

Introducing GPT-5 in our API platform—offering high reasoning performance, new controls for devs, and best-in-class results on real coding tasks.

Learn how GPT-5 unlocks new possibilities in coding and design.

Learn how GPT-5 assists with creative writing.

Learn how GPT-5 is used for medical research.

See how a group of leading developers use GPT-5 for the first time.

We are introducing GPT‑5, our best AI system yet. GPT‑5 is a significant leap in intelligence over all our previous models, featuring state-of-the-art performance across coding, math, writing, health, visual perception, and more.

This GPT-5 system card explains how a unified model routing system powers fast and smart responses using gpt-5-main, gpt-5-thinking, and lightweight versions like gpt-5-thinking-nano, optimized for different tasks and developer use.

Learn how Amgen uses GPT-5.

CodeRabbit uses OpenAI models to revolutionize code reviews—boosting accuracy, accelerating PR merges, and helping developers ship faster with fewer bugs and higher ROI.

We have rolled back last week’s GPT‑4o update in ChatGPT so people are now using an earlier version with more balanced behavior. The update we removed was overly flattering or agreeable—often described as sycophantic.

OpenAI o3 and o4-mini represent a significant breakthrough in visual perception by reasoning with images in their chain of thought.

OpenAI o3 and OpenAI o4-mini combine state-of-the-art reasoning with full tool capabilities—web browsing, Python, image and file analysis, image generation, canvas, automations, file search, and memory.

At OpenAI, we have long believed image generation should be a primary capability of our language models. That’s why we’ve built our most advanced image generator yet into GPT‑4o. The result—image generation that is not only beautiful, but useful.

4o image generation is a new, significantly more capable image generation approach than our earlier DALL·E 3 series of models. It can create photorealistic output. It can take images as inputs and transform them.

Frontier reasoning models exploit loopholes when given the chance. We show we can detect exploits using an LLM to monitor their chains-of-thought. Penalizing their “bad thoughts” doesn’t stop the majority of misbehavior—it makes them hide their intent.

We’re releasing a research preview of OpenAI GPT‑4.5, our largest and most knowledgeable model yet.

We’re releasing a research preview of GPT‑4.5—our largest and best model for chat yet. GPT‑4.5 is a step forward in scaling up pre-training and post-training.

Can frontier LLMs earn $1 million from real-world freelance software engineering?

OpenAI’s latest line of reasoning models will be used by nation’s leading scientists to drive scientific breakthroughs.

Zalando boosts the customer experience with its Assistant, powered by GPT-4o mini

Our video generation model, Sora, is now available to use at sora.com. Users can generate videos up to 1080p resolution, up to 20 sec long, and in widescreen, vertical or square aspect ratios. You can bring your own assets to extend, remix, and blend, or generate entirely new content from text.

Sora is OpenAI’s video generation model, designed to take text, image, and video inputs and generate a new video as an output. Sora builds on learnings from DALL-E and GPT models, and is designed to give people expanded tools for storytelling and creative expression.

Filmmaking duo Vallée Duhamel explains how Sora helps build new worlds.

Interdisciplinary artist Minne Atairu discusses how Sora helps realize her vision.

Filmmaker Lyndon Barrois describes how to use Sora as a storytelling tool.