A Public Benefit Corporation informing society about the frontier of physical AI.
- Tweets 125
- Following 26
- Followers 2,153
- Likes 578
Robocurve PBC retweeted
Even really simple things like this require strong knowledge of physics and good reactivity
Dream-1 by @dreamlabs_ai is the first model to complete "Stand Up Bottles", a difficult robotics task that frontier LLMs struggle with.
2
28
2,054
Robocurve PBC retweeted
Tomorrow's panel at AI Safety Night SF: "What does it mean for AI to be safe enough to act in the physical world?"
🎙️ @chooi_jeq (@robocurve)
🎙️ @apagajewski (@pantographPBC)
🎙️ @jerryhuang01 (@RoboticsCtr)
Mod: Christina Hsu (Intuit)
Posters are still open!
AI Safety Night SF is back 🔥
Last time: 150+ people, sold out.
What’s new:
- Theme: From Multimodal to Physical AI
- A poster session where YOU present
Tue, Oct 20 · San Francisco
Poster submissions close Oct 17
Hosting with our CFO, @Skye_Kim_25
Links below 👇
3
3
20
1,024
Robocurve PBC retweeted
incredibly excited for the adoption trajectory of @robocurve.
the physical AI benchmarks and evals market will note a similar trajectory to the software AI benchmarks (from generalized tasks to economic tasks). as this continues to accelerate, and as deployment continues to be a crucial focus of my companies - a standard benchmark is needed to assess the performance of models in the real world.
Check out the full 120 eval rollouts done by @robocurve . Full transparency is the only way to move the industry forward.
1
3
18
1,485
Robocurve PBC retweeted
Stand Up Bottles was one of the hardest tasks across all previous models, so it was surreal seeing Dream-1 being able to do this.
We will need more tasks to see what robotics models struggle with.
Dream-1 by @dreamlabs_ai is the first model to complete "Stand Up Bottles", a difficult robotics task that frontier LLMs struggle with.
2
1
37
1,876
Robocurve PBC retweeted
Introducing Embodied First Exam:
One line of code to run coding agents in 1,090 embodied tasks across 31 simulation suites and 44 robots.
Can your agent go far beyond pick-and-place:
Puzzles with spatial constraints;
Olympic sports with multi-robot collab;
Real-time eval, where the world won't pause;
Dynamic, agile, whole-body control;
AutoResearch with sim2sim
Try your own agent:
harbor run -d embodied-first-exam -a <your_own_harness> -m <your_own_model>
23
97
555
94,423
Robocurve PBC retweeted
Check out the full 120 eval rollouts done by @robocurve . Full transparency is the only way to move the industry forward.
We release all 120 Dream-1 robot runs, with full transcripts and videos: robocurve.org/all-runs/?lead…
1
5
62
7,330
Robocurve PBC retweeted
This team is one big reason I believe robotics is closer than it may appear
Dream-1 by @dreamlabs_ai beats every LLM in robotics while being faster than all of them. Dream-1 scores 1.3x higher than GPT-6 Astra Ultrafast while being 1.7x faster.
5
7
70
5,683
Robocurve PBC retweeted
In six days, the robotics SOTA jumped 11 points from Gemini 4 Argon (48%) to Dream-1 (59%). Dream-1 is the first model to score above 50%.
Dream-1 by Dream Labs (@dreamlabs_ai) achieves SOTA in robotics, the first to score above 50% on RoboDojo-RC Tier 1. It beats all frontier LLMs.
8
25
191
11,798
Robocurve PBC retweeted
Ahead of launch, @robocurve evaluated Dream-1 independently. Result: a new Pareto frontier for robot control. 16x cheaper, 1.7x faster, and better performance (59 vs. 48 on RoboDojo-RC Tier 1) than the previous best alternative.
LLMs can plan. Robots still need a cheap, fast, and general System 1 to act.
5
14
133
28,324
Dream-1 by @dreamlabs_ai is the first model to complete "Stand Up Bottles", a difficult robotics task that frontier LLMs struggle with.
9
19
155
24,338
We run the evals with our open-source harness, Inspect Robots. It is designed to run any model, on any robot, on any task: inspect-robots.org
1
7
526
We're hiring for Member of Technical Staff (up to $300k base + equity). $5k for successful referrals. Apply or refer here: robocurve.org/careers/
6
416
Robocurve (@robocurve) is an independent Public Benefit Corporation evaluating frontier AI in the physical world. Learn more: robocurve.org
1
2
260
We run the evals with our open-source harness, Inspect Robots. It is designed to run any model, on any robot, on any task: inspect-robots.org
1
1
191
We're hiring for Member of Technical Staff (up to $300k base + equity). $5k for successful referrals. Apply or refer here: robocurve.org/careers/
2
162
Dream-1 by @dreamlabs_ai beats every LLM in robotics while being faster than all of them. Dream-1 scores 1.3x higher than GPT-6 Astra Ultrafast while being 1.7x faster.
1
13
60
8,509
We release all 120 Dream-1 robot runs, with full transcripts and videos: robocurve.org/all-runs/?lead…
1
3
564
Robocurve (@robocurve) is an independent Public Benefit Corporation evaluating frontier AI in the physical world. Learn more: robocurve.org
1
2
5
341
We run the evals with our open-source harness, Inspect Robots. It is designed to run any model, on any robot, on any task: inspect-robots.org
1
3
98
We're hiring for Member of Technical Staff (up to $300k base + equity). $5k for successful referrals. Apply or refer here: robocurve.org/careers/
2
86