AI News

Anthropic Releases Fable 5.2, NVIDIA Launches GR00T 2, Both Doubling Down on Robot Control on the Same Day

Cailian PressUpdated Nov 10 15:59

AI summary

Anthropic released its new-generation flagship models Fable 5.2 and Mythos 5.2. On the RoboDojo dual-arm simulated robot benchmark, Fable 5.2 achieved an average success rate of 64.9%, up from 27.6% for the previous generation Fable 5.1; over the same period, Claude Opus 5.5 reached 39.4% and GPT-6 Astra 29.8%. On the same day, NVIDIA launched GR00T 2, a robot model built on the Cosmos platform, and released the robot learning framework Isaac Lab 3.0; GR00T 2 can complete 40 of 50 tasks without specialized demonstrations, with an average success rate of 71% when one demonstration is provided per task, rising to 95% with five demonstrations.

Why read this

AI score 78

Two releases on the same day provide a comparison of success rates on control tasks between a general-purpose large model and a specialized robot model, offering a window into the diverging paths of embodied intelligence.

Topics

Tags

Body · AI translation

Cailian Press Financial News Agency

Open the APP

Anthropic and Nvidia Launch New Products on the Same Day: Are Robots Really About to "Grow Brains"?

STAR Market Daily, Zheng Yuanfang, Editor: Ji Sheng, 2026-11-10 15:03

① Anthropic's Fable 5.2, as a general-purpose large model, has demonstrated outstanding capabilities in robot control. ② Nvidia launched its new-generation robot model GR00T 2 and released a robot learning framework along with multiple technologies. ③ Brokerages noted that the entry of large-model vendors will accelerate progress in embodied models, and expectations for the industry's commercialization are continuing to rise.

STAR Market Daily, November 10 — On November 9 local time, Nvidia and Anthropic successively released new products, adding two more sparks to robotics.

First, Anthropic, which has been under intense scrutiny from the capital markets on the eve of its IPO, released two major new models: the new-generation flagship models Fable 5.2 and Mythos 5.2.

Among them, as a general-purpose large model, the capabilities Fable 5.2 demonstrated in robot control drew industry attention. Test data published by Anthropic show that on the RoboDojo dual-arm simulated robot benchmark, Fable 5.2 achieved an average success rate of 64.9%.

What level is 64.9%? The previous-generation large model Fable 5.1 scored only 27.6% on this test — in other words, a "0.1-generation" update raised the success rate by 37.3 percentage points; and in the same-period comparison for Fable 5.2, Claude Opus 5.5 scored 39.4%, while GPT-6 Astra scored 29.8%.

A higher success-rate score also means the model is stronger at judging the environment from camera images and arranging actions.

Moreover, Fable 5.2 also possesses generalization capability. Anthropic noted that "Fable 5.2 shows substantially improved capabilities in controlling physical robots, including tasks it was never explicitly trained on." In a demo video released by Anthropic, Fable 5.2 controlled a physical robotic arm to complete three tasks it had not been specially trained for: drawing patterns, unscrewing bottle caps, and solving a Rubik's cube.

Anthropic said Fable 5.2's performance in robot control "outperforms most AI models designed specifically for robot control."

This is also an important reason why Fable 5.2 has sparked a huge response — previously, the mainstream view held that robots needed more vertical large models designed specifically for embodied intelligence to effectively solve spatial understanding, action arrangement, and control, whereas Fable 5.2, as a general-purpose large model, while excelling in complex programming, scientific research tasks, computer operation, interdisciplinary reasoning, and chart recognition, has also achieved a major leap in robot control.

On the other side, on the same day, November 9 local time, Nvidia announced the launch of its all-new next-generation robot model GR00T 2, and released the robot learning framework Isaac Lab 3.0 along with multiple robot training and evaluation technologies.

Nvidia said GR00T 2 is an open frontier world model built on the Cosmos platform. The model is trained using robot and human data, including 100,000 hours of open-source first-person human-perspective data provided by Nvidia.

One of GR00T 2's main features is out-of-the-box task execution capability: of 50 tasks, 40 can be completed directly without task-specific demonstration training. In addition, the model supports robots learning skills by learning from human operation demonstrations, and can transfer learned skills to robots of different forms without re-demonstrating tasks for each robot form. According to Nvidia, with only 1 demonstration provided per task, GR00T 2 achieved an average success rate of 71% across 50 tasks; if 5 demonstrations are provided per task, the average success rate rises to 95%.

Previously, on November 5, Nvidia had just released four Cosmos 3 Nano open-weight models, covering multi-view scene generation, interactive simulation, and robot motion prediction; Apple's research team, on November 4, released the HALO humanoid robot autonomous control system, which, by synthesizing training data in 1,000 procedurally generated household scenes, enables robots to autonomously perform tasks such as walking, obstacle avoidance, sitting down, and pushing boxes in real environments without mapping.

It is worth mentioning that on November 4, Wang Xingxing, founder, chairman, and CEO of Unitree Robotics, said that 2027 will very likely become a critical turning point for the robotics industry or the development of robot models, and that robot capabilities are expected to see substantial improvement next year; the "ChatGPT moment" of embodied intelligence is already close at hand. Models are precisely the technical threshold for realizing this moment: AI models' understanding of physical contact and touch is still insufficient, and there is a misalignment between model inputs and outputs and the real world. He expects this problem to be resolved within the next one to three years.

An Orient Securities research report dated November 2 pointed out that the main reason the market had previously been rather pessimistic about the embodied intelligence industry was that it believed embodied models were distinct from large models, and worried that slow model progress by companies in the embodied industry would lead to limited application scenarios for embodied intelligence. However, as GPT-6Astra has already achieved good application practices in the embodied field, some investors have gradually come to believe that general large models are likewise expected to capture market opportunities in the embodied field, and that the entry of large-model makers will accelerate progress in embodied models; as a result, expectations for commercialization of the industry are continuing to rise.

The analysts further pointed out that large models have enormous potential as the embodied strategy layer, and that embodied hardware makers stand to benefit the most at the margin. As general large-model makers continue to explore the embodied intelligence field in the future, this is expected to drive accelerated application of embodied intelligence in logistics and industrial manufacturing, and makers with vertical-scenario know-how will have relatively higher barriers. It is expected that more large-model makers will accelerate their deployment in the embodied field in the future, which is expected to drive breakthroughs in more application scenarios and thereby accelerate breakthroughs in the industry's commercialization process, with robot complete-machine and component makers expected to benefit.

Artificial Intelligence

Robotics

Cailian Press statement: The content of this article is for reference only and does not constitute investment advice. Investors who act on it do so at their own risk.

Business Cooperation

Popular Unlocks

[Pan Zhong Bao · Reporters on the Front Line] First price increase in more than eight months, multiple companies in the industry collectively issue price increase notices

Reporters have learned that multiple companies in the industry have raised the sales prices of this type of product, marking the first price increase since late February this year and the first in more than eight months.

As AI large models continue to iterate, institutions are optimistic about the application opportunities brought by AI moving from model capabilities to scenario deployment next year. This company can provide customers with one-stop services, while another company ranks among the top in user numbers for its segmented business and has already applied large models to vertical scenarios.

[Pan Zhong Bao] Filling in the key to the perception loop of humanoid robots, demand for this type of hardware is expected to rise rapidly in the future, and this company has already deployed in related fields

Driving robots to evolve into "embodied intelligent partners," demand for this type of hardware is expected to rise rapidly in the future, companies are accelerating their entry, and this company has already deployed in related fields.

Related Reading

Memory duo vie for EUV lithography machines to support advanced DRAM capacity expansion, with single-company investment caps approaching 100 billion

STAR Market Daily Zhang Zhen 2026-11-10 09:56

① Samsung Electronics and SK Hynix are racing to order ASML's extreme ultraviolet lithography equipment, with each company's investment cap potentially reaching 20 trillion won; ② ASML warned in its October earnings report that planned supply will fail to meet demand in both 2027 and 2028; ③ Similar shortages are also occurring in the back-end equipment sector, with order backlogs at companies such as Hanmi Semiconductor and Amkor already lasting two to three years.

Nvidia CEO controversy continues to ferment! Wall Street heatedly discusses: just how important is Jensen Huang?

Cailian Press Huang Junzhi 2026-11-10 09:38

① On November 8 local time, Nvidia founder and CEO Jensen Huang sent an apology letter to company employees, admitting to an extramarital affair, after which Nvidia's board of directors launched an independent investigation; ② Analyses point out that Nvidia is highly dependent on its founder, that any leadership transition would be extremely risky, and that investors have almost no precedent for assessing Nvidia's value after losing Jensen Huang, though some analysts optimistically point out that he is not that important.

Undeterred by cooling market sentiment! Anthropic reportedly plans to sprint toward an IPO next week

Cailian Press Liu Rui 2026-11-10 08:55

① Anthropic plans to list on the U.S. stock market next Thursday or Friday, and is expected to list on Nasdaq under the ticker symbol ANT; ② Anthropic will hold roadshow meetings with core investors this week and is expected to submit documents early this week disclosing the IPO price range; ③ If the market environment deteriorates further later on, the company still has the possibility of postponing the IPO.

Source

Cailian Press