Heads-up for developers trying gpt-oss: performance and correctness can vary a bit across providers and runtimes right now due to implementation differences. We’re working with inference providers to make sure gpt-oss performs at its best everywhere, and we’d love your feedback!
@romainhuet
-
GPT-OSS Demo: Local Laptop Execution with Tools and Game Coding
By
–
Watch @DKundel and @zhuohan123 bring gpt-oss to life: running locally on a laptop, calling tools, and coding a game entirely offline with the Codex CLI. Fun to be behind the scenes for this one as the demos came together on Sunday!
-
GPT-OSS Reaches Number 1 Trending on Hugging Face
By
–
We’re already #1 trending on @huggingface with gpt-oss! Huge thanks to the open-source AI community for the love.
-
GPT-OSS Models Now Available in AI SDK Playground
By
–
Both gpt-oss-120b and gpt-oss-20b are now in @aisdk and you can test them nicely side-by-side in the AI SDK playground. ▲
-
GPT-OSS: Try Open Source Language Models Live Online
By
–
Try it live here → https://
gpt-oss.com You’ll also find instructions to download and run the models locally with your favorite tool — @huggingface
, @ollama
, or @lmstudio
. Unless you have a MacBook M3 Ultra or M4 Max with 128 GB RAM, start with the smaller gpt-oss-20b! -
Remarkable Performance of Compact Open-Source AI Models
By
–
What’s absolutely amazing to me is the size of these models:
• gpt-oss-120b runs on a high-end MacBook or on a single H100.
• gpt-oss-20b runs on consumer hardware. And yet, their evals are remarkable! Matching or exceeding o4-mini and o3-mini respectively in core benchmarks. -
Hugging Face Partnership Enables Local and Cloud Model Deployment
By
–
Partnering with @huggingface has been incredible. We also worked with @ollama
, @lmstudio
, @vllm_project
, so you can run models locally with your favorite tool on day one, and with many cloud and hardware partners so you can deploy it efficiently anywhere. -
Two AI Models Built for Agentic Tool Use and Structured Outputs
By
–
We learned how much features like agentic tool use and Structured Outputs matter to you. But also that one size wouldn’t fit all. That’s why we built two models, each designed for different needs and hardware, and applied our most rigorous safety testing.
-
Open Models Designed for Developers Through Community Input
By
–
From the start, we wanted to make these open models maximally useful for developers. So instead of guessing, we asked the developer community, hosting listening sessions in SF and online around the world to hear what was most important to you.
-
Lancement de deux nouveaux modèles de langage open-source
By
–
Today’s a big day! We have something really exciting to share with the open-source community.
— Romain Huet (@romainhuet) 5 août 2025
We’re launching two open-weight language models: gpt-oss-120b and gpt-oss-20b.
They’re incredible models, built for developers, trained for reasoning, efficiency, and real-world use.🧵 pic.twitter.com/ra5LsPgbGLToday’s a big day! We have something really exciting to share with the open-source community. We’re launching two open-weight language models: gpt-oss-120b and gpt-oss-20b. They’re incredible models, built for developers, trained for reasoning, efficiency, and real-world use.