Loading video...

Video Failed to Load

Go Home

Announcing Agent S3 — our most advanced computer-use agent yet — achieving 69.9% on OSWorld, just shy of the 72% human benchmark. Today also marks one year since Simular launched the first generation of Agent S. From 20% to 69.9% in just twelve months. What a year. Let that...

angli_ai's profile picture

Ang Li

1,536 subscribers

444,496 views • 9 months ago •via X (Twitter)

Science & TechnologyNews & PoliticsEducation
LIVE

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

🚀 Introducing 𝐀𝐠𝐞𝐧𝐭 𝐒3, the most advanced computer-use agent, now 𝐚𝐩𝐩𝐫𝐨𝐚𝐜𝐡𝐢𝐧𝐠 𝐡𝐮𝐦𝐚𝐧-𝐥𝐞𝐯𝐞𝐥 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞🧠💻 Just one year ago, Agent S scored ~20% on OSWorld: SOTA then, but far from human 72%. Today, Agent S3 reaches 6̳9̳.̳9̳%̳ (⬆10% over prior SOTA), nearly matching humans on computer use. Few imagined this gap could close so fast, but with steady progress and 𝐟𝐢𝐫𝐬𝐭-𝐩𝐫𝐢𝐧𝐜𝐢𝐩𝐥𝐞 simplicity, Simular did it in just one year. 🧪 The s̵e̵c̵r̵e̵t̵ 𝐨𝐩𝐞𝐧 𝐫𝐞𝐜𝐢𝐩𝐞: → 𝐁𝐞𝐡𝐚𝐯𝐢𝐨𝐫 𝐁𝐞𝐬𝐭-𝐨𝐟-𝐍 — scaling agents 𝘵𝘩𝘦 𝘳𝘪𝘨𝘩𝘵 𝘸𝘢𝘺 → Simpler design, stronger baseline → 100% 𝐨𝐩𝐞𝐧 𝐬𝐨𝐮𝐫𝐜𝐞 (as always) Learn more at & 👇
0:30

🚀 Introducing 𝐀𝐠𝐞𝐧𝐭 𝐒3, the most advanced computer-use agent, now 𝐚𝐩𝐩𝐫𝐨𝐚𝐜𝐡𝐢𝐧𝐠 𝐡𝐮𝐦𝐚𝐧-𝐥𝐞𝐯𝐞𝐥 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞🧠💻 Just one year ago, Agent S scored ~20% on OSWorld: SOTA then, but far from human 72%. Today, Agent S3 reaches 6̳9̳.̳9̳%̳ (⬆10% over prior SOTA), nearly matching humans on computer use. Few imagined this gap could close so fast, but with steady progress and 𝐟𝐢𝐫𝐬𝐭-𝐩𝐫𝐢𝐧𝐜𝐢𝐩𝐥𝐞 simplicity, Simular did it in just one year. 🧪 The s̵e̵c̵r̵e̵t̵ 𝐨𝐩𝐞𝐧 𝐫𝐞𝐜𝐢𝐩𝐞: → 𝐁𝐞𝐡𝐚𝐯𝐢𝐨𝐫 𝐁𝐞𝐬𝐭-𝐨𝐟-𝐍 — scaling agents 𝘵𝘩𝘦 𝘳𝘪𝘨𝘩𝘵 𝘸𝘢𝘺 → Simpler design, stronger baseline → 100% 𝐨𝐩𝐞𝐧 𝐬𝐨𝐮𝐫𝐜𝐞 (as always) Learn more at & 👇

Xin Eric Wang

913,690 views • 9 months ago

Today marks an important milestone in the history of Simular, the autonomous computer company. Our open source computer-use agent, 𝐀𝐠𝐞𝐧𝐭 𝐒, scored 72.6% on the OSWorld benchmark, surpassing the human baseline (72.36%) for the first time ever. This milestone matters because it shows AI can now use computers the way humans do, and, in many cases, do it better. This is a glimpse of a future where work becomes faster, more accessible, and more empowered for everyone. #AI #Automation #Simular #AgenticAI #ComputerUse
0:53

Today marks an important milestone in the history of Simular, the autonomous computer company. Our open source computer-use agent, 𝐀𝐠𝐞𝐧𝐭 𝐒, scored 72.6% on the OSWorld benchmark, surpassing the human baseline (72.36%) for the first time ever. This milestone matters because it shows AI can now use computers the way humans do, and, in many cases, do it better. This is a glimpse of a future where work becomes faster, more accessible, and more empowered for everyone. #AI #Automation #Simular #AgenticAI #ComputerUse

Simular

839,901 views • 6 months ago

🧠 Just dropped: Our first Chromia | Power to the Public AI Agent tutorial is LIVE! Learn to build AI Agent Brains right on the blockchain - with memory systems that mirror human cognition 👇🧵
0:50

🧠 Just dropped: Our first Chromia | Power to the Public AI Agent tutorial is LIVE! Learn to build AI Agent Brains right on the blockchain - with memory systems that mirror human cognition 👇🧵

superoo7

14,233 views • 1 year ago

Orchids is rebranding to Bud: the first agent with a full computer. Since our launch in September of last year, we’ve grown to well over 7 figures ARR in record time. We’ve simultaneously come to realize that an agent that builds apps is just the first step in a future where an agent does the majority of human work. Comment below to get early access or go to bud [dot] app.
0:51

Orchids is rebranding to Bud: the first agent with a full computer. Since our launch in September of last year, we’ve grown to well over 7 figures ARR in record time. We’ve simultaneously come to realize that an agent that builds apps is just the first step in a future where an agent does the majority of human work. Comment below to get early access or go to bud [dot] app.

Bud

123,288 views • 2 months ago

This AI doesn’t just automate. It uses your computer like a human clicks, types, navigates all from a simple prompt. Meet Simular Pro, the first production-grade computer-use agent. It runs thousands of steps, 24/7, so you don’t have to. How? Most agents flake. Simular Pro doesn’t. It combines two agents: 🧠 Neural (explores) ⚙️ Symbolic (executes) Every action = editable, deterministic code. Here’s how Simular works ↓
10:08

This AI doesn’t just automate. It uses your computer like a human clicks, types, navigates all from a simple prompt. Meet Simular Pro, the first production-grade computer-use agent. It runs thousands of steps, 24/7, so you don’t have to. How? Most agents flake. Simular Pro doesn’t. It combines two agents: 🧠 Neural (explores) ⚙️ Symbolic (executes) Every action = editable, deterministic code. Here’s how Simular works ↓

FELIX

76,011 views • 1 year ago

Introducing SIMA: the first generalist AI agent to follow natural-language instructions in a broad range of 3D virtual environments and video games. 🕹️ It can complete tasks similar to a human, and outperforms an agent trained in just one setting. 🧵
0:23

Introducing SIMA: the first generalist AI agent to follow natural-language instructions in a broad range of 3D virtual environments and video games. 🕹️ It can complete tasks similar to a human, and outperforms an agent trained in just one setting. 🧵

Google DeepMind

1,061,548 views • 2 years ago

We've been using Cua-Bench internally—and with customers—for the last few months to evaluate every computer-use agent we deploy. Today it's open-source. 15 public tasks, 40 variations, adapters for OSWorld and Windows Agent Arena. One CLI, self-hostable.
0:49

We've been using Cua-Bench internally—and with customers—for the last few months to evaluate every computer-use agent we deploy. Today it's open-source. 15 public tasks, 40 variations, adapters for OSWorld and Windows Agent Arena. One CLI, self-hostable.

Cua

87,427 views • 5 months ago

This AI agent just surpassed humans for the first time ever Simular open-sourced Agent S, and scored 72.6% on OSWorld bench vs 72.36% humans, automating your desktop and complex workflows 6 wild examples + how to try: 1. Turn research paper to X thread with assets
0:41

This AI agent just surpassed humans for the first time ever Simular open-sourced Agent S, and scored 72.6% on OSWorld bench vs 72.36% humans, automating your desktop and complex workflows 6 wild examples + how to try: 1. Turn research paper to X thread with assets

Alvaro Cintas

25,438 views • 6 months ago

Computer use agents are slow and brittle. The fix isn’t just stronger models, but also deploying them as multi-agent systems. MACU is a general Multi-Agent Computer Use framework that consistently lifts success rates by 3.4-25.5% and is up to 1.5x faster on long-horizon tasks.🧵
0:23

Computer use agents are slow and brittle. The fix isn’t just stronger models, but also deploying them as multi-agent systems. MACU is a general Multi-Agent Computer Use framework that consistently lifts success rates by 3.4-25.5% and is up to 1.5x faster on long-horizon tasks.🧵

Jing Yu Koh

27,760 views • 1 month ago

Introducing 21st Agent Elements Diffed edits. Plan approvals. Clarifying questions. Sub-agents. Tool cards. The pieces every "agent UI" lib skips. 25 shadcn /ui components built for the new generation of agents that use your computer. Open-source. Link in thread 🧵
0:28

Introducing 21st Agent Elements Diffed edits. Plan approvals. Clarifying questions. Sub-agents. Tool cards. The pieces every "agent UI" lib skips. 25 shadcn /ui components built for the new generation of agents that use your computer. Open-source. Link in thread 🧵

serafim

43,404 views • 2 months ago

For the first time, you can vibe-code any AI agent. Meet — Computer Human AI by Langbase ☕ 🔹Prompt: "make an agent that…" 🔹Sip: chai builds any AI agent 🔹Ship: every agent gets a UI 🤯 Like your on-demand AI Engineer. What will you s(h)ip today?
1:35

For the first time, you can vibe-code any AI agent. Meet — Computer Human AI by Langbase ☕ 🔹Prompt: "make an agent that…" 🔹Sip: chai builds any AI agent 🔹Ship: every agent gets a UI 🤯 Like your on-demand AI Engineer. What will you s(h)ip today?

Ahmad Awais

149,221 views • 1 year ago

It’s long been our prediction that 2025 will be the “year of the AI agent”. As the industry explores new and innovative ways to use agents, every company must adapt or die. Us too. Today we announce GameGPT’s AI Agent Platform. A full breakdown below. 🧵
0:25

It’s long been our prediction that 2025 will be the “year of the AI agent”. As the industry explores new and innovative ways to use agents, every company must adapt or die. Us too. Today we announce GameGPT’s AI Agent Platform. A full breakdown below. 🧵

GameGPT by PRISM

18,634 views • 1 year ago

One of our most exciting partnerships this year has been with @philaunion who launched the first official Philadelphia Union app this year, complete with a Theta-powered AI agent. We're excited for what's to come in 2026! Happy Holidays, folks! ☃️❄️
0:42

One of our most exciting partnerships this year has been with @philaunion who launched the first official Philadelphia Union app this year, complete with a Theta-powered AI agent. We're excited for what's to come in 2026! Happy Holidays, folks! ☃️❄️

Theta Network

11,494 views • 6 months ago

Open-source just won... Agent S2 is an AI agent that uses your computer like a human. → No fake demos → No terminal hallucinations → Just real results and it’s FREE. Here’s why this is the future of work: 👇
0:38

Open-source just won... Agent S2 is an AI agent that uses your computer like a human. → No fake demos → No terminal hallucinations → Just real results and it’s FREE. Here’s why this is the future of work: 👇

Shruti

206,646 views • 1 year ago

In December 2023, we launched our podcast. Just twelve months later, #NBA fans from 49 countries have downloaded our content. Here are the most popular episodes from the first two seasons… 🧵
1:35

In December 2023, we launched our podcast. Just twelve months later, #NBA fans from 49 countries have downloaded our content. Here are the most popular episodes from the first two seasons… 🧵

Basketball Now & Then

30,499 views • 1 year ago

🚨 BREAKING: Hugging Face just dropped a free AI agent that uses a computer like a human. It’s called Open Computer Agent, and it mimics real computer use. You can run it in your browser no install required. Here's everything you need to know:
0:29

🚨 BREAKING: Hugging Face just dropped a free AI agent that uses a computer like a human. It’s called Open Computer Agent, and it mimics real computer use. You can run it in your browser no install required. Here's everything you need to know:

Ihtesham Haider

245,607 views • 1 year ago

Today we're announcing the Vibecode CLI Your OpenClaw (or any agent) can build and deploy up to 20 apps in parallel in a single prompt. Your agent will communicate with Vibecode to build your app(s). Frontend, Backend, Database, Deployments, APIs. Works with OpenClaw, Claude Code, Perplexity Computer, Manus, and any agent with access to a terminal. Just add this to your prompt: "Please use vibecode . dev CLI to build these apps"
2:20

Today we're announcing the Vibecode CLI Your OpenClaw (or any agent) can build and deploy up to 20 apps in parallel in a single prompt. Your agent will communicate with Vibecode to build your app(s). Frontend, Backend, Database, Deployments, APIs. Works with OpenClaw, Claude Code, Perplexity Computer, Manus, and any agent with access to a terminal. Just add this to your prompt: "Please use vibecode . dev CLI to build these apps"

vibecode.dev

114,468 views • 3 months ago

Today’s Agent Spotlight: Image Generation Agent, powered by the HiDream-I1 model via Chutes on ASI:One. ⚡ Generate stunning images: fast, secure, in-chat - using just text. Photorealistic, cartoon, artistic or surreal: just describe what you want. ASI:One takes care of the rest. Try now on Link to agent: Watch 👇
1:54

Today’s Agent Spotlight: Image Generation Agent, powered by the HiDream-I1 model via Chutes on ASI:One. ⚡ Generate stunning images: fast, secure, in-chat - using just text. Photorealistic, cartoon, artistic or surreal: just describe what you want. ASI:One takes care of the rest. Try now on Link to agent: Watch 👇

Fetch.ai

52,148 views • 11 months ago

2024 was a big year for us. We launched the first agentic framework, expanded our ecosystem, and saw growth like never before. Everything is falling into place, from new users and creators to the most advanced Agent platform. Thanks to everyone who’s been part of this journey. Let’s take a look back at what we achieved in 2024:
0:52

2024 was a big year for us. We launched the first agentic framework, expanded our ecosystem, and saw growth like never before. Everything is falling into place, from new users and creators to the most advanced Agent platform. Thanks to everyone who’s been part of this journey. Let’s take a look back at what we achieved in 2024:

MyShell.AI

775,351 views • 1 year ago

PokéLLMon A Human-Parity Agent for Pokémon Battles with Large Language Models paper page: introduce POKE´LLMON, the first LLMembodied agent that achieves human-parity performance in tactical battle games, as demonstrated in Pokemon battles. The design of ´ POKE´LLMON incorporates three key strategies: (i) In-context reinforcement learning that instantly consumes textbased feedback derived from battles to iteratively refine the policy; (ii) Knowledge-augmented generation that retrieves external knowledge to counteract hallucination and enables the agent to act timely and properly; (iii) Consistent action generation to mitigate the panic switching phenomenon when the agent faces a powerful opponent and wants to elude the battle. We show that online battles against human demonstrates POKE´LLMON’s human-like battle strategies and just-in-time decision making, achieving 49% of win rate in the Ladder competitions and 56% of win rate in the invited battles.
1:00

PokéLLMon A Human-Parity Agent for Pokémon Battles with Large Language Models paper page: introduce POKE´LLMON, the first LLMembodied agent that achieves human-parity performance in tactical battle games, as demonstrated in Pokemon battles. The design of ´ POKE´LLMON incorporates three key strategies: (i) In-context reinforcement learning that instantly consumes textbased feedback derived from battles to iteratively refine the policy; (ii) Knowledge-augmented generation that retrieves external knowledge to counteract hallucination and enables the agent to act timely and properly; (iii) Consistent action generation to mitigate the panic switching phenomenon when the agent faces a powerful opponent and wants to elude the battle. We show that online battles against human demonstrates POKE´LLMON’s human-like battle strategies and just-in-time decision making, achieving 49% of win rate in the Ladder competitions and 56% of win rate in the invited battles.

AK

102,163 views • 2 years ago