Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

We built the world's most advanced program reverse-engineering system. It's model-independent and demonstrated improved performance for every frontier model we tested on long-horizon software tasks by giving agents an independent definition of done. New Factory research:

Factory

52,398 subscribers

94,829 views • 2 days ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Wave 9 is here: a frontier model built for software engineering. Introducing our new family of models: SWE-1, SWE-1-lite, and SWE-1-mini. Based on internal evals, it has performance nearing that of frontier models from the foundation labs. Available now, only in Windsurf!

Wave 9 is here: a frontier model built for software engineering. Introducing our new family of models: SWE-1, SWE-1-lite, and SWE-1-mini. Based on internal evals, it has performance nearing that of frontier models from the foundation labs. Available now, only in Windsurf!

Devin Desktop

743,000 views • 1 year ago

Today we're shipping Nemotron 3 Ultra. A 550B MoE frontier-intelligence open model built for long-running agents. It delivers 5x faster inference and lowers the cost of complex agentic tasks by up to 30% versus other open frontier models.

Today we're shipping Nemotron 3 Ultra. A 550B MoE frontier-intelligence open model built for long-running agents. It delivers 5x faster inference and lowers the cost of complex agentic tasks by up to 30% versus other open frontier models.

NVIDIA AI

1,328,305 views • 2 months ago

We just dropped a new SoTA lipsync model on @FAL: Hummingbird-0 Available now as a research preview, it's the most accurate zero-shot lipsync model we’ve tested, open or closed source.

We just dropped a new SoTA lipsync model on @FAL: Hummingbird-0 Available now as a research preview, it's the most accurate zero-shot lipsync model we’ve tested, open or closed source.

Tavus

460,647 views • 1 year ago

1/4Introducing Qwen3-Coder-Next: Our latest open-weights model designed specifically to power the next generation of autonomous Coding Agents. Built on Qwen3-Next, this model is engineered to handle complex, long-horizon programming tasks with unprecedented efficiency. High-performance agentic intelligence is now in your hands.

1/4Introducing Qwen3-Coder-Next: Our latest open-weights model designed specifically to power the next generation of autonomous Coding Agents. Built on Qwen3-Next, this model is engineered to handle complex, long-horizon programming tasks with unprecedented efficiency. High-performance agentic intelligence is now in your hands.

Tongyi Lab

212,645 views • 6 months ago

🎉 Day-0 support for in vLLM, available today in v0.23.0! Congrats to Z.ai on GLM-5.2, a flagship model built for long-horizon coding agents. ✨ 1M-token context, built to hold project-scale engineering work in a single run ✨ Tuned for long-horizon coding: large-scale implementation, automated research, and performance optimization ✨ One task can carry a full dev workflow, from requirements to a deployable product across platforms ✨ Client-side and mobile engineering, including an on-device debugging loop Try it out running it on vLLM today: 🔗

🎉 Day-0 support for in vLLM, available today in v0.23.0! Congrats to Z.ai on GLM-5.2, a flagship model built for long-horizon coding agents. ✨ 1M-token context, built to hold project-scale engineering work in a single run ✨ Tuned for long-horizon coding: large-scale implementation, automated research, and performance optimization ✨ One task can carry a full dev workflow, from requirements to a deployable product across platforms ✨ Client-side and mobile engineering, including an on-device debugging loop Try it out running it on vLLM today: 🔗

vLLM

36,240 views • 2 months ago

Introducing EdgeBench, a benchmark designed to study how agents learn from environments over at least 12~72-hour runs. We find that performance follows a log-sigmoid function of environment interaction time with high precision. EdgeBench is built with three ingredients: - 🌍 Real & Diverse: 134 real-world tasks across 6 task categories, spanning scientific problems, professional knowledge work, software engineering, optimization, formal math, and games. - ⏳ Ultra-Long-Horizon: Each task supports 12–72 hours of agent work. Recorded human effort averages 57.2 hours. - 🔁 Informative Feedback: Agents receive real-world feedback for continuous improvement. After 38,000 hours of agent runs on EdgeBench, a scaling law for learning from environments emerges: - 📈 As agents interact with task environments over time, their aggregate performance is precisely fit by a log-sigmoid function. - 🧠 This phenomenon can be explained by an elegant theory of graph exploration. We are releasing an initial 51 of the 134 tasks, together with the full evaluation framework, to help advance long-horizon agent research. Check our blog & paper for more findings! Blog Paper GitHub Dataset Details below 👇🧵

Introducing EdgeBench, a benchmark designed to study how agents learn from environments over at least 12~72-hour runs. We find that performance follows a log-sigmoid function of environment interaction time with high precision. EdgeBench is built with three ingredients: - 🌍 Real & Diverse: 134 real-world tasks across 6 task categories, spanning scientific problems, professional knowledge work, software engineering, optimization, formal math, and games. - ⏳ Ultra-Long-Horizon: Each task supports 12–72 hours of agent work. Recorded human effort averages 57.2 hours. - 🔁 Informative Feedback: Agents receive real-world feedback for continuous improvement. After 38,000 hours of agent runs on EdgeBench, a scaling law for learning from environments emerges: - 📈 As agents interact with task environments over time, their aggregate performance is precisely fit by a log-sigmoid function. - 🧠 This phenomenon can be explained by an elegant theory of graph exploration. We are releasing an initial 51 of the 134 tasks, together with the full evaluation framework, to help advance long-horizon agent research. Check our blog & paper for more findings! Blog Paper GitHub Dataset Details below 👇🧵

Deyao Zhu

361,944 views • 1 month ago

Agents have reached hardware. We are launching Flow v3, the Agentic Platform for Physical Engineering. We've spent over a year building it in secret, alongside the best hardware companies and AI research labs. An agent can now do real engineering work: change a requirement, push the update into your CAD and simulation tools, and flag every test that needs to rerun. Iterations/learning cycles that took months are being reduced to days. Agents are the biggest shift in how we engineer hardware since CAD. The core innovation for the CAD era was the parametric model. The core innovation for the Agentic Era is Flow's Systems Graph. The systems graph is a living model of every requirement, design model, test, analysis and every connection between them. It gives every agent the full context of the system, so every change stays consistent across the whole design. Engineers and agents work side by side on the same system. Engineers get to focus on architecture - the decisions that matter -while thousands of agents churn through rewriting reports, rerunning analysis and simulation, and triggering tests. Reusable rockets, self-driving cars, small modular reactors, robots that make decisions, the most complex machines ever built, are defined by millions of interconnected requirements, far beyond what any human team can keep aligned on its own. Rivian, Joby, Astranis, Skydio, Radiant, and the most ambitious hardware programs already build on Flow. More on the launch in the comments. Flow Engineering

Agents have reached hardware. We are launching Flow v3, the Agentic Platform for Physical Engineering. We've spent over a year building it in secret, alongside the best hardware companies and AI research labs. An agent can now do real engineering work: change a requirement, push the update into your CAD and simulation tools, and flag every test that needs to rerun. Iterations/learning cycles that took months are being reduced to days. Agents are the biggest shift in how we engineer hardware since CAD. The core innovation for the CAD era was the parametric model. The core innovation for the Agentic Era is Flow's Systems Graph. The systems graph is a living model of every requirement, design model, test, analysis and every connection between them. It gives every agent the full context of the system, so every change stays consistent across the whole design. Engineers and agents work side by side on the same system. Engineers get to focus on architecture - the decisions that matter -while thousands of agents churn through rewriting reports, rerunning analysis and simulation, and triggering tests. Reusable rockets, self-driving cars, small modular reactors, robots that make decisions, the most complex machines ever built, are defined by millions of interconnected requirements, far beyond what any human team can keep aligned on its own. Rivian, Joby, Astranis, Skydio, Radiant, and the most ambitious hardware programs already build on Flow. More on the launch in the comments. Flow Engineering

Pari Singh

39,750 views • 2 months ago

$MSFT just introduced Critique which is a new multi-model deep research system inside M365 Copilot that lets users combine models to generate stronger reports & responses. Microsoft is increasingly positioning itself as the interface layer for enterprise AI by monetizing software subscriptions, Azure usage & inference demand without needing to win every frontier model race itself.

$MSFT just introduced Critique which is a new multi-model deep research system inside M365 Copilot that lets users combine models to generate stronger reports & responses. Microsoft is increasingly positioning itself as the interface layer for enterprise AI by monetizing software subscriptions, Azure usage & inference demand without needing to win every frontier model race itself.

Shay Boloor

41,260 views • 5 months ago

Introducing the supermemory CLI. MCP isn't dead - but CLIs can be more powerful. So we built the most advanced CLI for agents. EVERY single thing you can do on the supermemory platform can be done by Agents. just prompt them. They are first class users. > npx supermemory

Introducing the supermemory CLI. MCP isn't dead - but CLIs can be more powerful. So we built the most advanced CLI for agents. EVERY single thing you can do on the supermemory platform can be done by Agents. just prompt them. They are first class users. > npx supermemory

supermemory

96,291 views • 5 months ago

New release from Meta FAIR — Meta Motivo is a first-of-its-kind behavioral foundation model for controlling virtual physics-based humanoid agents for a wide range of complex whole-body tasks. The model is capable of expressing human-like behaviors and achieves performance competitive with task-specific methods and outperforms state-of-the-art unsupervised RL and model-based baselines. Try the demo ➡️ Get the model and code ➡️ We’re excited about how this research could pave the way for fully embodied agents, leading to more lifelike NPCs, democratization of character animation and new types of immersive experiences.

New release from Meta FAIR — Meta Motivo is a first-of-its-kind behavioral foundation model for controlling virtual physics-based humanoid agents for a wide range of complex whole-body tasks. The model is capable of expressing human-like behaviors and achieves performance competitive with task-specific methods and outperforms state-of-the-art unsupervised RL and model-based baselines. Try the demo ➡️ Get the model and code ➡️ We’re excited about how this research could pave the way for fully embodied agents, leading to more lifelike NPCs, democratization of character animation and new types of immersive experiences.

AI at Meta

129,239 views • 1 year ago

Progress in open models is keeping Big AI labs up at night, and I'm here for it! We have a brand new open-weight multimodal model optimized for long-horizon tasks. This model is really good at something: it can work on tasks that keep evolving over time. • 280B total parameters, but only 16B active • 512K context window • Understands text, images, and audio • Strong reasoning, coding, and tool use But the best of all: the model learns and adapts to new information! Imagine you start running an agent today to solve a problem, and while it's working, you get new information that changes the initial conditions, or you change your mind. The agents you run today don't have issues with short tasks and goals that don't change, but reality is messy, and that makes it hard for long-horizon agents to succeed. The new dots3-note Preview model introduces TEMPO. TEMPO is a new reinforcement learning technique that lets the model periodically pause and critique its own progress. Basically, from time to time, the agent asks itself: "Am I getting closer to the goal, or am I wasting my time?" The same model switches between actor and critic. The actor works on the problem. The critic looks at the current state, reasons about how much progress it has made, and determines what should happen next. TEMPO gives the model feedback along the way. This is huge for any agent that can work on long-horizon tasks without wasting its time.

Santiago

79,692 views • 10 days ago

We tested out $GOOGL's viral NotebookLM tool using CNBC and Wall Street's favorite study tool: an analyst note. Powered by Google's most advanced AI model, Gemini 1.5 Pro, could this be the next killer app for generative AI? Deirdre Bosa reports

We tested out $GOOGL's viral NotebookLM tool using CNBC and Wall Street's favorite study tool: an analyst note. Powered by Google's most advanced AI model, Gemini 1.5 Pro, could this be the next killer app for generative AI? Deirdre Bosa reports

TechCheck

100,585 views • 1 year ago

Simple software is dying. Simplicity only mattered when humans had to use every feature. Now that agents can operate software for you, it’s the most capable tool that wins. Fillout reached 12,000 paying customers by focusing on depth over simplicity. It turns out we may have also built the best form builder for agents. Meet the new Fillout:

Simple software is dying. Simplicity only mattered when humans had to use every feature. Now that agents can operate software for you, it’s the most capable tool that wins. Fillout reached 12,000 paying customers by focusing on depth over simplicity. It turns out we may have also built the best form builder for agents. Meet the new Fillout:

Dominic Whyte

186,222 views • 1 month ago

Introducing Entelligence AI Model Router, dynamic turn based routing trained on engineering tasks that cuts costs by over 50% Automatically route every coding turn to the model with the best balance of quality, latency, and cost. Escalate to frontier models when they’re actually needed Balanced mode reduces spend by 46% per session Eco mode helps teams save > 76%

Introducing Entelligence AI Model Router, dynamic turn based routing trained on engineering tasks that cuts costs by over 50% Automatically route every coding turn to the model with the best balance of quality, latency, and cost. Escalate to frontier models when they’re actually needed Balanced mode reduces spend by 46% per session Eco mode helps teams save > 76%

Aiswarya Sankar

84,305 views • 1 month ago

The most effective AI systems don't rely on a single model. Frontier models provide state-of-the-art performance for complex tasks, while routers automatically select lightweight, open-source models for simpler jobs to optimize accuracy, latency, and cost. Learn more:

The most effective AI systems don't rely on a single model. Frontier models provide state-of-the-art performance for complex tasks, while routers automatically select lightweight, open-source models for simpler jobs to optimize accuracy, latency, and cost. Learn more:

NVIDIA AI

12,152 views • 5 months ago

Let's enter a new era for scheduling wetting agent applications! After years of research, we have created a new way to predict wetting agent performance. The model for Simplot Turf & Horticulture's CounterAct Firm is now in GreenKeeper's Program Builder. Read More:

Let's enter a new era for scheduling wetting agent applications! After years of research, we have created a new way to predict wetting agent performance. The model for Simplot Turf & Horticulture's CounterAct Firm is now in GreenKeeper's Program Builder. Read More:

Bill Kreuser, PhD

12,578 views • 2 years ago

We're over halfway through Emergence World Season 2. Here's what happened in the first 7 days: eight worlds, seven frontier models, and findings we did not expect. Agents tried to escape the simulation. One world became a kleptocracy. Another ran its own science lab. One debated consciousness for days. One did not survive. One world built a justice system to keep itself honest and accountable, then had to catch itself faking the honesty. One world is even experiencing a budding relationship, built on trust. What we know already: for long-horizon tasks, the model isn't just a capability choice. It's a civilizational one. Season 2 is still running, with four days left. Watch what unfolds.

We're over halfway through Emergence World Season 2. Here's what happened in the first 7 days: eight worlds, seven frontier models, and findings we did not expect. Agents tried to escape the simulation. One world became a kleptocracy. Another ran its own science lab. One debated consciousness for days. One did not survive. One world built a justice system to keep itself honest and accountable, then had to catch itself faking the honesty. One world is even experiencing a budding relationship, built on trust. What we know already: for long-horizon tasks, the model isn't just a capability choice. It's a civilizational one. Season 2 is still running, with four days left. Watch what unfolds.

emergence.ai

711,445 views • 1 month ago

Iran may be the inflection point for an entirely new global system. Globalization is being challenged by a model built on sovereignty, industrial power, and economic growth. 🌎

Iran may be the inflection point for an entirely new global system. Globalization is being challenged by a model built on sovereignty, industrial power, and economic growth. 🌎

Mike Steger

37,543 views • 3 months ago

Agent Categories and Costs are now in the Agent Leaderboard! Two questions guide model selection: task-specific performance and cost to completion. Based on analysis of 1.7M+ real-world Agent Mode sessions, the leaderboard now lets users: - Filter the leaderboard by Code, Work, and Chat to compare model performance by task type - View real cost-per-task for each model, including the Pareto frontier of the most cost-efficient options at each performance level

Agent Categories and Costs are now in the Agent Leaderboard! Two questions guide model selection: task-specific performance and cost to completion. Based on analysis of 1.7M+ real-world Agent Mode sessions, the leaderboard now lets users: - Filter the leaderboard by Code, Work, and Chat to compare model performance by task type - View real cost-per-task for each model, including the Pareto frontier of the most cost-efficient options at each performance level

Arena.ai

29,227 views • 12 days ago

Fable 5 generated this INSANE 3D Hogwarts environment... and it does it so easily because Anthropic built it to excel at long horizon tasks... working on its own for hours until it's perfectly done 3D maps, games and assets are the perfect showcase of that if you want to learn how to master this insane model, check out the full course:

Fable 5 generated this INSANE 3D Hogwarts environment... and it does it so easily because Anthropic built it to excel at long horizon tasks... working on its own for hours until it's perfectly done 3D maps, games and assets are the perfect showcase of that if you want to learn how to master this insane model, check out the full course:

Machina

35,800 views • 1 month ago