Loading video...

Video Failed to Load

Go Home

OpenAI shows how gpt-oss can autonomously beat 2048 using reinforcement learning (RL). Training was done locally with Unsloth on NVIDIA DGX Spark. You can also do it free on Colab. 🦥 OpenAI DevDay notebook:

UnslothAI's profile picture

Unsloth AI

68,825 subscribers

98,847 views • 8 months ago •via X (Twitter)

Science & Technology
LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

We teamed up with NVIDIA and Matthew Berman to teach you how to do Reinforcement Learning! Learn about: - RL environments, reward functions & reward hacking - Training OpenAI gpt-oss to automatically solve 2048 - Local Windows training with NVIDIA_AI_PC RTX GPUs - How RLVR (verifiable rewards) works - How to interpret RL metrics like KL Divergence Full video tutorial:
5:13

We teamed up with NVIDIA and Matthew Berman to teach you how to do Reinforcement Learning! Learn about: - RL environments, reward functions & reward hacking - Training OpenAI gpt-oss to automatically solve 2048 - Local Windows training with NVIDIA_AI_PC RTX GPUs - How RLVR (verifiable rewards) works - How to interpret RL metrics like KL Divergence Full video tutorial:

Unsloth AI

54,179 views • 6 months ago

"I don’t have a GPU" is no longer an excuse 🤯 You can now train LLMs directly in VS Code using a free Google Colab runtime. → Connect any fine-tuning notebook to Colab → Train locally or on a free cloud GPU → Works with Unsloth
0:38

"I don’t have a GPU" is no longer an excuse 🤯 You can now train LLMs directly in VS Code using a free Google Colab runtime. → Connect any fine-tuning notebook to Colab → Train locally or on a free cloud GPU → Works with Unsloth

Alvaro Cintas

86,217 views • 4 months ago

Thank you Google for demoing how to fine-tune Gemma 3 with Unsloth, free on Colab! 🦥 #GoogleIO
4:09

Thank you Google for demoing how to fine-tune Gemma 3 with Unsloth, free on Colab! 🦥 #GoogleIO

Unsloth AI

49,143 views • 1 year ago

OpenAI has finally released GPT OSS 20B!! This is probably the most powerful model you can run locally on a laptop. → Only 3.6B active parameters → Adjustable "Reasoning Effort" → Works with 16GB of RAM! Quick steps to use it 100% free and offline:
0:29

OpenAI has finally released GPT OSS 20B!! This is probably the most powerful model you can run locally on a laptop. → Only 3.6B active parameters → Adjustable "Reasoning Effort" → Works with 16GB of RAM! Quick steps to use it 100% free and offline:

Paul Couvert

595,522 views • 11 months ago

gpt-oss with built-in websearch in ollama running fully locally on my mac with Exa for search shoutout to @DKundel OpenAI Developers for making it so easy :)
1:01

gpt-oss with built-in websearch in ollama running fully locally on my mac with Exa for search shoutout to @DKundel OpenAI Developers for making it so easy :)

parth

19,862 views • 10 months ago

GPT-5.3-Codex-Spark is now in research preview. You can just build things—faster.
0:47

GPT-5.3-Codex-Spark is now in research preview. You can just build things—faster.

OpenAI

1,523,881 views • 4 months ago

openai gpt-oss 20b model speed on rtx 4090 windows ollama
0:40

openai gpt-oss 20b model speed on rtx 4090 windows ollama

Mert Cobanov

465,600 views • 11 months ago

Just a little bit faster than everyone else on OpenAI GPT-OSS-120B. > 10 times faster than the leading NVIDIA GPU provider
0:23

Just a little bit faster than everyone else on OpenAI GPT-OSS-120B. > 10 times faster than the leading NVIDIA GPU provider

Andrew Feldman

124,175 views • 10 months ago

When NVIDIA CEO Jensen Huang presented the DGX Spark, then he recalled how, back in 2016, he unveiled the world’s first AI supercomputer, the DGX-1, and how Elon Musk, then at OpenAI, became its first customer. The DGX Spark is far more advanced, delivering five times the computational power of the DGX-1.
1:58

When NVIDIA CEO Jensen Huang presented the DGX Spark, then he recalled how, back in 2016, he unveiled the world’s first AI supercomputer, the DGX-1, and how Elon Musk, then at OpenAI, became its first customer. The DGX Spark is far more advanced, delivering five times the computational power of the DGX-1.

SMX 🇺🇸

13,312,184 views • 7 months ago

OpenAI just released GPT-OSS: An Open Source Language Model on Hugging Face Open source meaning: 💸 Free 🔒 Private 🔧 Customizable
0:47

OpenAI just released GPT-OSS: An Open Source Language Model on Hugging Face Open source meaning: 💸 Free 🔒 Private 🔧 Customizable

dylan

21,568 views • 11 months ago

Reinforcement Learning is the future tense of intelligence. Echo is how it scales. Echo is Gradient’s distributed RL framework, running on everyday consumer devices. From its early experiments, Echo powered a 30B Sokoban model that outperformed DeepSeek-R1 and GPT-OSS-120B.
0:55

Reinforcement Learning is the future tense of intelligence. Echo is how it scales. Echo is Gradient’s distributed RL framework, running on everyday consumer devices. From its early experiments, Echo powered a 30B Sokoban model that outperformed DeepSeek-R1 and GPT-OSS-120B.

Gradient

279,022 views • 10 months ago

Unsloth AI and NVIDIA are Revolutionizing Local LLM Fine-Tuning: From RTX Desktops to DGX Spark Fine-tune popular AI models faster with Unsloth on NVIDIA RTX AI PCs such as GeForce RTX desktops and laptops to RTX PRO workstations and the new DGX Spark to build personalized assistants for coding, creative work, and complex agentic workflows. The landscape of modern AI is shifting. We are moving away from a total reliance on massive, generalized cloud models and entering the era of local, agentic AI. Whether it is tuning a chatbot to handle hyper-specific product support or building a personal assistant that manages intricate schedules, the potential for generative AI on local hardware is boundless. However, developers face a persistent bottleneck: How do you get a Small Language Model (SLM) to punch above its weight class and respond with high accuracy for specialized tasks? The answer is Fine-Tuning, and the tool of choice is Unsloth. Unsloth provides an easy and high-speed method to customize models. Optimized for efficient, low-memory training on NVIDIA GPUs, Unsloth scales effortlessly from GeForce RTX desktops and laptop all the way to the DGX Spark, the world’s smallest AI supercomputer...... Full analysis: NVIDIA NVIDIA AI NVIDIA AIDev NVIDIAnewsroom Unsloth AI Unsloth
0:40

Unsloth AI and NVIDIA are Revolutionizing Local LLM Fine-Tuning: From RTX Desktops to DGX Spark Fine-tune popular AI models faster with Unsloth on NVIDIA RTX AI PCs such as GeForce RTX desktops and laptops to RTX PRO workstations and the new DGX Spark to build personalized assistants for coding, creative work, and complex agentic workflows. The landscape of modern AI is shifting. We are moving away from a total reliance on massive, generalized cloud models and entering the era of local, agentic AI. Whether it is tuning a chatbot to handle hyper-specific product support or building a personal assistant that manages intricate schedules, the potential for generative AI on local hardware is boundless. However, developers face a persistent bottleneck: How do you get a Small Language Model (SLM) to punch above its weight class and respond with high accuracy for specialized tasks? The answer is Fine-Tuning, and the tool of choice is Unsloth. Unsloth provides an easy and high-speed method to customize models. Optimized for efficient, low-memory training on NVIDIA GPUs, Unsloth scales effortlessly from GeForce RTX desktops and laptop all the way to the DGX Spark, the world’s smallest AI supercomputer...... Full analysis: NVIDIA NVIDIA AI NVIDIA AIDev NVIDIAnewsroom Unsloth AI Unsloth

Marktechpost AI Dev News ⚡

31,551 views • 6 months ago

This afternoon I picked up a new Nvidia DGX Spark computer with the goal of trying to run Gemma 4 31b (4bit) on it locally as a server. Just 1.5 hours later, it’s working! Using Open WebUI on my MacBook as the interface and it’s connecting to my DGX Spark running as a Gemma 4 server.
0:46

This afternoon I picked up a new Nvidia DGX Spark computer with the goal of trying to run Gemma 4 31b (4bit) on it locally as a server. Just 1.5 hours later, it’s working! Using Open WebUI on my MacBook as the interface and it’s connecting to my DGX Spark running as a Gemma 4 server.

Dave Lee

138,489 views • 2 months ago

GPT-5 makes building easy, Pietro Schirano shows how.
2:03

GPT-5 makes building easy, Pietro Schirano shows how.

OpenAI

355,935 views • 10 months ago

Jensen Huang hand-delivers the new NVIDIA DGX Spark to Elon Musk at Starbase 2016: Jensen hand-delivered the first DGX-1 to Elon at OpenAI Now: Nearly a decade later, he’s delivering the next generation at Starbase • DGX-1 (2016): 3,000 Watts • DGX Spark (2025): 40 Watts
1:10

Jensen Huang hand-delivers the new NVIDIA DGX Spark to Elon Musk at Starbase 2016: Jensen hand-delivered the first DGX-1 to Elon at OpenAI Now: Nearly a decade later, he’s delivering the next generation at Starbase • DGX-1 (2016): 3,000 Watts • DGX Spark (2025): 40 Watts

X Freeze

20,547 views • 5 months ago

How GPT-5 thinks, with OpenAI VP of Research Jerry Tworek 00:00 - Intro 01:01 - What Reasoning Actually Means in AI 02:32 - Chain of Thought: Models Thinking in Words 05:25 - How Models Decide How Long to Think 07:24 - Evolution from o1 to o3 to GPT-5 11:00 - The Road to OpenAI: Growing up in Poland, Dropping out of School, Trading 20:32 - Working on Robotics and Rubik's Cube Solving 23:02 - A Day in the Life: Talking to Researchers 24:06 - How Research Priorities Are Determined 26:53 - OpenAI's Culture of Transparency 29:32 - Balancing Research with Shipping Fast 31:52 - Using OpenAI's Own Tools Daily 32:43 - Pre-Training Plus RL: The Modern AI Stack 35:10 - Reinforcement Learning 101: Training Dogs 40:17 - The Evolution of Deep Reinforcement Learning 42:09 - When GPT-4 Seemed Underwhelming at First 45:39 - How RLHF Made GPT-4 Actually Useful 48:02 - Unsupervised vs Supervised Learning 49:59 - GRPO and How DeepSeek Accelerated US Research 53:05 - What It Takes to Scale Reinforcement Learning 55:36 - Agentic AI and Long-Horizon Thinking 59:19 - Alignment as an RL Problem 1:01:11 - Winning ICPC World Finals Without Specific Training 1:05:53 - Applying RL Beyond Math and Coding 1:09:15 - The Path from Here to AGI 1:12:23 - Pure RL vs Language Models
1:16:04

How GPT-5 thinks, with OpenAI VP of Research Jerry Tworek 00:00 - Intro 01:01 - What Reasoning Actually Means in AI 02:32 - Chain of Thought: Models Thinking in Words 05:25 - How Models Decide How Long to Think 07:24 - Evolution from o1 to o3 to GPT-5 11:00 - The Road to OpenAI: Growing up in Poland, Dropping out of School, Trading 20:32 - Working on Robotics and Rubik's Cube Solving 23:02 - A Day in the Life: Talking to Researchers 24:06 - How Research Priorities Are Determined 26:53 - OpenAI's Culture of Transparency 29:32 - Balancing Research with Shipping Fast 31:52 - Using OpenAI's Own Tools Daily 32:43 - Pre-Training Plus RL: The Modern AI Stack 35:10 - Reinforcement Learning 101: Training Dogs 40:17 - The Evolution of Deep Reinforcement Learning 42:09 - When GPT-4 Seemed Underwhelming at First 45:39 - How RLHF Made GPT-4 Actually Useful 48:02 - Unsupervised vs Supervised Learning 49:59 - GRPO and How DeepSeek Accelerated US Research 53:05 - What It Takes to Scale Reinforcement Learning 55:36 - Agentic AI and Long-Horizon Thinking 59:19 - Alignment as an RL Problem 1:01:11 - Winning ICPC World Finals Without Specific Training 1:05:53 - Applying RL Beyond Math and Coding 1:09:15 - The Path from Here to AGI 1:12:23 - Pure RL vs Language Models

Matt Turck

451,229 views • 8 months ago

nvidia allows you to use many llm models on their website. to use it programatically you can either create a phone verified nvidia nim account (heavily limited, often slow and NOT AVAILABLE IN TURKEY) or my wrapper app which bypasses this measures. > how it works? - it spawns and invisible safari webview of nvidia page - it solves the captcha using js (because page uses really soft version of hCaptcha) - because it's easy, it can create a pool of tokens - then spawns a local openai-anthropic compatible proxy server on port 2289 so you can use it with any tool, locally.
0:20

nvidia allows you to use many llm models on their website. to use it programatically you can either create a phone verified nvidia nim account (heavily limited, often slow and NOT AVAILABLE IN TURKEY) or my wrapper app which bypasses this measures. > how it works? - it spawns and invisible safari webview of nvidia page - it solves the captcha using js (because page uses really soft version of hCaptcha) - because it's easy, it can create a pool of tokens - then spawns a local openai-anthropic compatible proxy server on port 2289 so you can use it with any tool, locally.

Batuhan Katırcı

54,298 views • 2 months ago

BOOOOM! You can now run OpenAI gpt-oss 20B natively in Colaboratory T4 for FREE! 🔥 Powered by Transformers ⚡ The setup takes a bit since everything is bleeding edge, but once done it should work as expected Link to our cookbook in comments 👇
1:15

BOOOOM! You can now run OpenAI gpt-oss 20B natively in Colaboratory T4 for FREE! 🔥 Powered by Transformers ⚡ The setup takes a bit since everything is bleeding edge, but once done it should work as expected Link to our cookbook in comments 👇

Vaibhav (VB) Srivastav

37,474 views • 11 months ago

Codex can now help you build AI apps and agents faster with OpenAI APIs using the OpenAI Developers plugin.
0:40

Codex can now help you build AI apps and agents faster with OpenAI APIs using the OpenAI Developers plugin.

OpenAI Developers

318,434 views • 1 month ago

Roman Huet, Head of Developer Experience at OpenAI, shows a live demo of how GPT-4o can interact with the world through a webcam, understanding what it sees and reads
2:23

Roman Huet, Head of Developer Experience at OpenAI, shows a live demo of how GPT-4o can interact with the world through a webcam, understanding what it sees and reads

Tsarathustra

225,946 views • 2 years ago