OpenAI Releases Two 'Open' AI Models 👀

Welcome to another edition of Horizon AI,

The AI startup returns to its open-source roots with the release of two “open-weight” large language models, which are free to download and can be customized by developers.

Let’s take a closer look!

Read Time: 4.5 min

Here's what's new today in the Horizon AI

• OpenAI Launches Two ‘Open’ AI Reasoning Models

• Google’s New AI Model Can Create Interactive 3D Worlds

• AI Tutorial: How to run the new OpenAI model locally

• AI Tools to check out

• AI Findings/Resources

• The Latest in AI and Tech 💡

AI News

OPENAI

OpenAI Launches Two ‘Open’ AI Reasoning Models

OpenAI is launching a new open-weight model, GPT-OSS, which is free to download, fully customizable, and capable of running on a laptop.

Details:

• The model comes in two variants: 120-billion-parameter and 20-billion-parameter versions. The first can run on a single NVIDIA GPU, and the smaller version can run on just 16GB of VRAM.

• Both models are available on Hugging Face and GitHub, under the Apache 2.0 license, which allows them to be freely modified and used for commercial purposes.

• According to OpenAI, the gpt-oss-120b model performs on par with the o4-mini model on most key benchmarks, while the gpt-oss-20b is comparable to o3-mini — and even surpasses it in some cases.

• The models are multilingual and designed for use in agentic workflows, including tool use, web search, and Python code execution, while also offering adjustable reasoning levels (low, medium, high).

The startup announced these models earlier this year, and after a couple of delays, they’re finally here, marking their first “open” release in more than six years since the release of GPT-2 in 2019.

Source

TOGETHER WITH SKEJ

Meet your new assistant (who happens to be AI).

Skej is your new scheduling assistant. Whether it’s a coffee intro, a client check-in, or a last-minute reschedule, Skej is on it. Just CC Skej on your emails, and it takes care of everything:

• Customize your assistant name, email, and personality

• Easily manages time zones and locales

• Works with Google, Outlook, Zoom, Slack, and Teams

• Skej works 24/7 in over 100 languages

No apps to download or new tools to learn. You talk to Skej just like a real assistant, and Skej just… works! It’s like having a super-organized co-worker with you all day.

Get 10 meetings free

GOOGLE

Google’s New AI Model Can Create Interactive 3D Worlds

Google DeepMind has announced Genie 3, a new "world model" that can create interactive 3D environments in real time. This means you can type in a prompt and watch as a whole world appears, ready for you or an AI agent to explore and interact with.

Details:

• Genie 3 is a big step up from earlier versions. Now, users can move around in these AI-made worlds for a few minutes at a time, instead of just seconds.

• The model also remembers where things are, so if you look away and then back, details like wall paint or chalkboard writing stay consistent.

• Users can explore these dynamic virtual worlds at 720p resolution and 24 frames per second.

• They can also use prompts to change things on the fly, like the weather or which characters appear.

• There are still limits, like how you interact with the world and how text appears, but Google says it’s working on improvements.

Right now, Genie 3 is only available to a small group of researchers and creators. Google wants to study how people use it and make sure it’s safe before letting more people try it.

Source

AI Tutorial

How to run the new OpenAI model locally

Here's how you can run the new GPT OSS 20B locally on your laptop — 100% free and offline:

1. Download LM Studio for your OS (Windows, macOS, or Linux).

2. Type "GPT OSS 20B" into the search bar.

3. Download the recommended model.

4. Open the Chat tab and load the model from the top bar.

5. Set the "Reasoning Effort" and start chatting.

AI Tools to check out

⭐ TestSprite: Turns your messy code into working software.

📊 Infogrammy: Turn your data into stunning infographics.

👀 Tavus: Build AI humans that can see, hear, interact and take action – in real time.

✍ Rytr: AI writing assistant for short-form content, accessible anywhere you write.

📝 PrompTessor: Evaluate, improve, and optimize your AI prompts.

AI Findings/Resources

👀 9 creative ways to use ChatGPT that you probably haven’t tried yet

💼 AI is impacting the labor market, young tech workers

🤔 Behind the Curtain: What does AI owe you?

💡 Why open-source AI became an American national priority

👉 5 Prompts to help you get started with ChatGPT agent mode

The latest in AI and Tech

Anthropic releases powerful Opus 4.1 model with improved coding & agent capabilities

The company released an upgraded version of its flagship AI model, achieving new performance heights in software engineering tasks — scoring 74.5% on SWE-bench Verified, a benchmark for real-world coding problems — surpassing OpenAI’s o3 model at 69.1% and Google’s Gemini 2.5 Pro at 67.2%.

Anthropic describes Claude 4.1 as a hybrid reasoning model designed to handle both instant outputs and extended thinking. Developers can fine-tune “thinking budgets” via the API to balance cost and performance.

Alibaba releases a new AI image generation model

Alibaba’s “Qwen Team” just released Qwen-Image, a new image generation model that stands out due to its emphasis on rendering text accurately within visuals, allowing users to create content like movie posters, presentation slides, and stylized infographics.

The model is available under the Apache 2.0 license on Qwen Chat, Hugging Face, and GitHub.

Gemini app can now create custom storybooks, comics, more

Google rolled out a new feature for the Gemini app called “Storybook” that lets you generate “personalized, illustrated storybooks complete with read-aloud narration” just by describing what you want.

You can also upload photos and other images for Gemini to reference.

OpenAI models are available on AWS for the first time

The two new open-weight reasoning models will be available as a model choice with Amazon AI services Bedrock and SageMaker AI.

The partnership comes as a surprise, given that until now, AWS has been best known as a major host and a financial backer of Anthropic’s Claude, one of the main competitors of OpenAI.

That’s a wrap!

Thanks for sticking with us to the end! Let’s stay connected on LinkedIn and Twitter.

Not subscribed yet? Sign up here and send it to a colleague or friend!

See you in our next edition!

Gina 👩🏻‍💻