Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

We did a comparison study on where a robot's intelligence should live by using the same model (OpenAI GPT6-Astra), same bimanual YAM station, same LEGO pick-and-place task, all run on SE3' real-world remote eval stack: 1. Frozen code-as-policy (built on NVIDIA ENPIRE components): 6/20 placements 2. Agentic control +...

paigeinsf's profile picture

Paige

4,468 subscribers

30,002 Aufrufe • vor 1 Monat •via X (Twitter)

Wissenschaft & TechnologieBildung
LIVE

Anya Rossi• Live Now

Private livecam show

6 Kommentare

Profilbild von Paige
Paigevor 1 Monat

Here’s how the two approaches worked: 1. Code-as-policy: Astra wrote and tested a Python policy, which we then froze for evaluation. The policy ran without further model calls. 2. Agentic control: Each episode started with a fresh Astra session that read camera snapshots and issued commands through a control harness.

Profilbild von Paige
Paigevor 1 Monat

Getting hold of the brick wasn’t enough. What mattered was robustness and recovery. The code-as-policy lifted the brick in 14/20 episodes but completed only 6 placements. In the other eight, it stopped before reaching the release pose or held the brick without releasing it. Agentic control completed 18/20 placements. Nine required grasp retries, including one that recovered from a drop.

Profilbild von Paige
Paigevor 1 Monat

The agent recovered from its own misses: 9 of its 18 completions needed more than one grasp. In episode 8, it missed five times, lifted on the sixth, dropped the brick, reacquired it and released at 419s. The code-as-policy never made a second grasp attempt.

Profilbild von Paige
Paigevor 1 Monat

Agentic control completed more placements, but successful runs took longer. Among successful episodes, median time to release was 131 seconds for code-as-policy versus 175 seconds for agentic control. Agentic control also had more time to recover: up to 600 seconds, compared with 180 seconds for code-as-policy. The timing gap reflects more than model reasoning alone.

Profilbild von Paige
Paigevor 1 Monat

Full report and all 40 episodes on video: Reach out if you have a policy or an idea you’d like to evaluate on real robots!

Profilbild von catman
catmanvor 1 Monat

@OpenAI @se3labs @nvidia Keeping the model in the loop lets it adjust each movement from live feedback, rather than letting errors compound through a fixed sequence of actions.

Ähnliche Videos

GPT-6 Astra vs Claude Opus 5.5 (both on High) Same prompt. A storm inside a bottle, built from scratch in Blender, all code, zero assets. - Astra: 24 min, ~$6 - Opus: 82 min, ~$18 Was Opus worth 3x the price?
0:25

GPT-6 Astra vs Claude Opus 5.5 (both on High) Same prompt. A storm inside a bottle, built from scratch in Blender, all code, zero assets. - Astra: 24 min, ~$6 - Opus: 82 min, ~$18 Was Opus worth 3x the price?

Mr Tactical

42,309 Aufrufe • vor 25 Tagen

Astra vs Fable 5 : Jump is same like opus 4.8 to fable 5 in quality side by side comparison on same prompts , this is why i am super hyped for openai < astra is the path to the bell of AGI iykyk>
1:23

Astra vs Fable 5 : Jump is same like opus 4.8 to fable 5 in quality side by side comparison on same prompts , this is why i am super hyped for openai < astra is the path to the bell of AGI iykyk>

Chetaslua

296,201 Aufrufe • vor 2 Monaten

π0.5 vs. MolmoAct2. Same task. Same object placements. Same training data regimen. Precise spatial placement lets us compare models side by side on any matching rollout in our Reality Check benchmark. Someone jokingly called it “Model Smash.” 👇 1/3
0:40

π0.5 vs. MolmoAct2. Same task. Same object placements. Same training data regimen. Precise spatial placement lets us compare models side by side on any matching rollout in our Reality Check benchmark. Someone jokingly called it “Model Smash.” 👇 1/3

Nicolas Keller

13,260 Aufrufe • vor 1 Monat

SITUATION EXPLAINED: Grok 4.7 beats GPT-6 Astra on real-world tasks, and it's 5x cheaper. • Same price as 4.6, $2 input and $6 output per million tokens, well under Sol, Fable, and Astra • A new larger base model, which Elon has put at 2.1 trillion parameters, up from 1.5 trillion, with extra training on SpaceX data • Trained to natively understand the Grok Bot harness, the same move Anthropic made with Claude Code • Leads on electrical engineering and Harvey's legal benchmark, beats Astra Max on GDPval, trails on software engineering • Elon Musk: "Grok 4.7 places SpaceXAI as third after Anthropic and OpenAI for agentic coding. When factoring in that Grok is significantly faster and lower cost, it's a great choice for your everyday workhorse" Theo Jaffee: "SpaceX has a huge amount of data on real world hardware problems, real world engineering. So I bet Grok models are going to be better at rocketry engineering than any of the other models out there."
7:29

SITUATION EXPLAINED: Grok 4.7 beats GPT-6 Astra on real-world tasks, and it's 5x cheaper. • Same price as 4.6, $2 input and $6 output per million tokens, well under Sol, Fable, and Astra • A new larger base model, which Elon has put at 2.1 trillion parameters, up from 1.5 trillion, with extra training on SpaceX data • Trained to natively understand the Grok Bot harness, the same move Anthropic made with Claude Code • Leads on electrical engineering and Harvey's legal benchmark, beats Astra Max on GDPval, trails on software engineering • Elon Musk: "Grok 4.7 places SpaceXAI as third after Anthropic and OpenAI for agentic coding. When factoring in that Grok is significantly faster and lower cost, it's a great choice for your everyday workhorse" Theo Jaffee: "SpaceX has a huge amount of data on real world hardware problems, real world engineering. So I bet Grok models are going to be better at rocketry engineering than any of the other models out there."

MTS

18,190 Aufrufe • vor 1 Monat

Sachin Katti says OpenAI had Astra optimize its own inference on NVIDIA's Rubin chip, adding another 2x in 72 hours "It's a testament to NVIDIA's hardware that out of the box we got a 3x improvement in throughput. We literally took the same code that we had optimized for Blackwell, ran it on Rubin, got 3x improvement." "Then, can we actually make Astra run agents using Astra as the base model, go optimize this inference of Astra itself on Rubin? It's a recursive loop. The model is figuring out how to optimize its own inference serving on Rubin chips, and over 72 hours the model produced another 2x improvement on the throughput." "Things that would take weeks, if not months, to optimize in the past, we now are getting models that are very capable of actually producing over a weekend." _______ Link and more from Sachin Katti's conversation:
0:48

Sachin Katti says OpenAI had Astra optimize its own inference on NVIDIA's Rubin chip, adding another 2x in 72 hours "It's a testament to NVIDIA's hardware that out of the box we got a 3x improvement in throughput. We literally took the same code that we had optimized for Blackwell, ran it on Rubin, got 3x improvement." "Then, can we actually make Astra run agents using Astra as the base model, go optimize this inference of Astra itself on Rubin? It's a recursive loop. The model is figuring out how to optimize its own inference serving on Rubin chips, and over 72 hours the model produced another 2x improvement on the throughput." "Things that would take weeks, if not months, to optimize in the past, we now are getting models that are very capable of actually producing over a weekend." _______ Link and more from Sachin Katti's conversation:

Fireside Alpha

40,013 Aufrufe • vor 1 Monat

Racism has no place where we all live on the same planet and breathe the same air.🥺🍿
6:05

Racism has no place where we all live on the same planet and breathe the same air.🥺🍿

Oyinkansola🎀💐

30,130 Aufrufe • vor 4 Monaten

1. Sam Altman on Will All Startups Be Killed by OpenAI There are 2 strategies to build on AI right now: Assume the model won’t improve. Assume OpenAI will stay on the same growth trajectory. 95% should bet on #2. We have a mission to improve the model, so we will steamroll you.
0:57

1. Sam Altman on Will All Startups Be Killed by OpenAI There are 2 strategies to build on AI right now: Assume the model won’t improve. Assume OpenAI will stay on the same growth trajectory. 95% should bet on #2. We have a mission to improve the model, so we will steamroll you.

Harry Stebbings

212,291 Aufrufe • vor 2 Jahren

Liquid AI's head of post-training explained how they built a small model that runs on-device under 1 GB in 20 minutes - better than $2500 small-model bootcamps. pick LFM2.5 base -> on-policy preference alignment -> agentic reinforcement learning -> curriculum training -> iterative model merging -> ship a 1B model that reliably calls tools on your phone. That loop is why frontier small models now beat 70B models on the tasks that actually matter. LFM2.5 + on-policy DPO + agentic RL + curriculum training + iterative merging - that's the stack. Watch and save it, then run a 1B agent on your phone tonight.
19:11

Liquid AI's head of post-training explained how they built a small model that runs on-device under 1 GB in 20 minutes - better than $2500 small-model bootcamps. pick LFM2.5 base -> on-policy preference alignment -> agentic reinforcement learning -> curriculum training -> iterative model merging -> ship a 1B model that reliably calls tools on your phone. That loop is why frontier small models now beat 70B models on the tasks that actually matter. LFM2.5 + on-policy DPO + agentic RL + curriculum training + iterative merging - that's the stack. Watch and save it, then run a 1B agent on your phone tonight.

h100envy

87,975 Aufrufe • vor 3 Monaten

GPT-6.1 Sol vs GPT-6 Astra on Higgsfield. Both models got the same task: build a 3D game prototype - with GPT-6.1 Sol completing the same workflow 4x faster at one-fifth the cost.
1:00

GPT-6.1 Sol vs GPT-6 Astra on Higgsfield. Both models got the same task: build a 3D game prototype - with GPT-6.1 Sol completing the same workflow 4x faster at one-fifth the cost.

Higgsfield AI 🧩

54,208 Aufrufe • vor 1 Monat

MY FIRST GAME OUTPUT WITH GPT-6 ASTRA IT ONE-SHOTTED THIS CALL OF DUTY GAME IN 30 MINUTES!!! GPT-6 ASTRA IS THE FASTEST MODEL ON THE MARKET. THE SAME PROMPT TOOK FABLE 5.1 ALMOST 5 HRS AND IT CONSUMED A FRACTION OF THE TOKENS FABLE 5.1 DID.
0:39

MY FIRST GAME OUTPUT WITH GPT-6 ASTRA IT ONE-SHOTTED THIS CALL OF DUTY GAME IN 30 MINUTES!!! GPT-6 ASTRA IS THE FASTEST MODEL ON THE MARKET. THE SAME PROMPT TOOK FABLE 5.1 ALMOST 5 HRS AND IT CONSUMED A FRACTION OF THE TOKENS FABLE 5.1 DID.

hiraeth

40,821 Aufrufe • vor 2 Monaten

Imagine a world where retail traders operate at the same efficiency as institutional desks where you don’t need 50 analysts to have the same edge Introducing TrueNorth, the world’s first agentic brokerage built on Hyperliquid 🧵
0:45

Imagine a world where retail traders operate at the same efficiency as institutional desks where you don’t need 50 analysts to have the same edge Introducing TrueNorth, the world’s first agentic brokerage built on Hyperliquid 🧵

TrueNorth AI

248,697 Aufrufe • vor 5 Monaten

Someone tested Union-Alpha, the new stealth model on OpenRouter, with their usual shooter prompt and it built a full playable FPS. The menu alone gives it away: huge title, orange arrow button, dark 3D scene, the exact same look as GPT-6 Astra. Feels like OpenAI is behind.
1:45

Someone tested Union-Alpha, the new stealth model on OpenRouter, with their usual shooter prompt and it built a full playable FPS. The menu alone gives it away: huge title, orange arrow button, dark 3D scene, the exact same look as GPT-6 Astra. Feels like OpenAI is behind.

ALICE ⚡ | AI Growth Mentor

108,338 Aufrufe • vor 1 Monat

NVIDIA just revealed MotionBricks, a new AI model trained on more than 350,000 motion clips. What’s really impressive is that the same model can animate a digital character and control a humanoid robot in the real world.
0:29

NVIDIA just revealed MotionBricks, a new AI model trained on more than 350,000 motion clips. What’s really impressive is that the same model can animate a digital character and control a humanoid robot in the real world.

Justin Ryan

85,343 Aufrufe • vor 3 Monaten

Muse Wonder might beat Astra on benchmarks, but it's nowhere close to Astra on taste. Same prompt: - Wonder (xhigh) = $2.42 in 20 min - Astra (low) = $2.19 in 7 min Benchmarks don't tell you everything 😬
0:21

Muse Wonder might beat Astra on benchmarks, but it's nowhere close to Astra on taste. Same prompt: - Wonder (xhigh) = $2.42 in 20 min - Astra (low) = $2.19 in 7 min Benchmarks don't tell you everything 😬

Anshu

23,253 Aufrufe • vor 1 Monat

Opus 5.5 vs GPT 6 astra tested both with the same prompt at their highest reasoning settings, using the same devin harness: > opus took 2 hours 10 mins and cost $75 > astra took 1 hour 25 mins and cost $61 opus 5.5 attention to details is on point, it took time to make everything perfect as i wanted but still super slow and ended up costing way more then astra which one did better here?
1:12

Opus 5.5 vs GPT 6 astra tested both with the same prompt at their highest reasoning settings, using the same devin harness: > opus took 2 hours 10 mins and cost $75 > astra took 1 hour 25 mins and cost $61 opus 5.5 attention to details is on point, it took time to make everything perfect as i wanted but still super slow and ended up costing way more then astra which one did better here?

J A Z I I

55,525 Aufrufe • vor 1 Monat

Up to 8x faster, and now you know why ⚡️@openai GPT-6 Astra Ultrafast runs on NVIDIA Read the story:
0:45

Up to 8x faster, and now you know why ⚡️@openai GPT-6 Astra Ultrafast runs on NVIDIA Read the story:

NVIDIA

490,661 Aufrufe • vor 1 Monat

I asked GPT-6 Astra to recreate famous scenes from Zelda OoT. It did it, using the same graphics as original game.
0:24

I asked GPT-6 Astra to recreate famous scenes from Zelda OoT. It did it, using the same graphics as original game.

Kevin Ngo

14,459 Aufrufe • vor 2 Monaten

"You have a data center full of geniuses and now you make them use the same desktop and move the same mouse cursor?" Some raw thoughts on the GPT-6 Astra launch.
4:46

"You have a data center full of geniuses and now you make them use the same desktop and move the same mouse cursor?" Some raw thoughts on the GPT-6 Astra launch.

Thorsten Ball

17,588 Aufrufe • vor 2 Monaten

Same prompt for all 4 tests, 100% code generated animation. GPT 6.1 Sol vs GPT 6 Astra vs Sonnet 5.5 vs Opus 5.5 Despite the huge price difference between the final results, if this was a blind test, we would not even say that Sol and Astra are on the same level as Sonnet and Opus. On price, Sol was much cheaper than all other models, and maybe even better than Astra quality wise. On quality alone, Opus and Sonnet were above (by a lot), including richer details, camera movements, effects, being more fluid and engaging overall. And if we consider Price and Quality, Opus is the overall winner, costing 12$ vs 18$ from Sonnet, finishing in 1h and 16m vs 1h and 36m.
1:06

Same prompt for all 4 tests, 100% code generated animation. GPT 6.1 Sol vs GPT 6 Astra vs Sonnet 5.5 vs Opus 5.5 Despite the huge price difference between the final results, if this was a blind test, we would not even say that Sol and Astra are on the same level as Sonnet and Opus. On price, Sol was much cheaper than all other models, and maybe even better than Astra quality wise. On quality alone, Opus and Sonnet were above (by a lot), including richer details, camera movements, effects, being more fluid and engaging overall. And if we consider Price and Quality, Opus is the overall winner, costing 12$ vs 18$ from Sonnet, finishing in 1h and 16m vs 1h and 36m.

Daniel Zambrini

35,664 Aufrufe • vor 1 Monat

Everyone was calling Fable5.1 the ceiling. Then GPT-6 Astra rendered the same scene. The bottom half is the file both models were handed. Crash dummies, no materials, no light, no faces. The top half is what each one built from it. GPT-6 Astra goes first. Fable5.1 goes second. Same seats, same sunroof, same autumn road. Watch the belt on the shoulder and the light through the glass. One model lit a car. The other remembered a drive. You will not need the label to know the second it switches.
0:26

Everyone was calling Fable5.1 the ceiling. Then GPT-6 Astra rendered the same scene. The bottom half is the file both models were handed. Crash dummies, no materials, no light, no faces. The top half is what each one built from it. GPT-6 Astra goes first. Fable5.1 goes second. Same seats, same sunroof, same autumn road. Watch the belt on the shoulder and the light through the glass. One model lit a car. The other remembered a drive. You will not need the label to know the second it switches.

AiMind

44,681 Aufrufe • vor 1 Monat