SenseTime has open-sourced the SenseNova-RoboRSI framework and a 27B embodied model, while also releasing an empirical study on embodied intelligence for GPT-6-Astra. The study covers 12 simulation benchmarks and 8 real robot platforms. Under a unified protocol (ICL plus up to 10x interaction budget), GPT-6-Astra achieved SOTA on all 12 benchmarks, with RoboDojo's success rate rising from 27.6% at 1x budget to 81.9% at 10x, and DexJoCo reaching 87.3%, compared with the previous SOTA of 40.3%. The study also notes that simple tasks such as putting a pen into a pen holder cost about $55 in inference, and that real robots overheat and wait too long; the conclusion is that GPT-6-Astra has essentially crossed the threshold of embodied capability, but practical-grade embodied intelligence has not yet arrived. The open-source framework packages each benchmark as an independent Docker image and provides environment interaction and scoring through a unified HTTP API; the 27B model raises the success rate on RoboCasa365 from 26.0% to 46.5%, reduces task token overhead by 73%, and costs about 1/50 of GPT-6-Astra per completed task.
SenseTime has open-sourced the SenseNova-RoboRSI framework and a 27B embodied model, while also releasing an empirical study on embodied intelligence for GPT-6-Astra. The study covers 12 simulation benchmarks and 8 real robot platforms. Under a unified protocol (ICL plus up to 10x interaction budget), GPT-6-Astra achieved SOTA on all 12 benchmarks, with RoboDojo's success rate rising from 27.6% at 1x budget to 81.9% at 10x, and DexJoCo reaching 87.3%, compared with the previous SOTA of 40.3%. Th
Researchers from Microsoft and City University of Hong Kong have built Taste-Bench to evaluate the long-term decision-making capabilities of AI agents on long tasks. The benchmark contains 502 binary-choice questions. According to The Decoder, the best-performing model, GPT-5.6 Sol, achieved an average accuracy of only 59.7%, while GPT-5.5 reached 59.5%. When key clues only become apparent after a large amount of subsequent work, the average accuracy of all models drops to 21%, close to random guessing, and increasing the reasoning budget brings almost no improvement. The researchers also distilled the teacher model's judgments using Qwen3.6-27B as the base model; the student's accuracy on unseen tasks rose from 30% to 47.9%, and its success rate on 41 SWE-bench Pro tasks increased from 14.6% to 33.7%.
Researchers from Microsoft and City University of Hong Kong have built Taste-Bench to evaluate the long-term decision-making capabilities of AI agents on long tasks. The benchmark contains 502 binary-choice questions. According to The Decoder, the best-performing model, GPT-5.6 Sol, achieved an average accuracy of only 59.7%, while GPT-5.5 reached 59.5%. When key clues only become apparent after a large amount of subsequent work, the average accuracy of all models drops to 21%, close to random g
François Chollet commented that Gemini 4 Argon performs similarly to Astra on ARC-AGI-3, with Google returning to the frontier. On the same day, ARC Prize published the verified results for Gemini 4 Argon on the ARC-AGI series: ARC-AGI-3 scored 97.5% under the provider adapter harness at a cost of $4.8K; under the standard harness it scored 49.8% at a cost of $5.8K. ARC-AGI-2 scored 91.3% at $0.32 per task; ARC-AGI-1 scored 98.5% at $0.16 per task. ARC Prize said Google DeepMind is the second frontier AI lab to exceed 95% on ARC-AGI-3, and Argon's 97.5% is second only to GPT-6 Astra's 99.9%. The standard harness test could not complete all 55 games due to API issues; the high-reasoning setting completed only 19 games, so no score was published.
François Chollet commented that Gemini 4 Argon performs similarly to Astra on ARC-AGI-3, with Google returning to the frontier. On the same day, ARC Prize published the verified results for Gemini 4 Argon on the ARC-AGI series: ARC-AGI-3 scored 97.5% under the provider adapter harness at a cost of $4.8K; under the standard harness it scored 49.8% at a cost of $5.8K. ARC-AGI-2 scored 91.3% at $0.32 per task; ARC-AGI-1 scored 98.5% at $0.16 per task. ARC Prize said Google DeepMind is the second fr
Anthropic warns that small AI models may be capable of self-replicating on commercial IoT devices within a few months, thereby forming a distributed AI workload network that does not depend on data centers. Related experiments show that Qwen3.5-9B successfully installed and configured the software required to run Llama3.1-8B on eight of nine hardware types; GLM-5.2 reached all eight nodes of a controlled simulated network in four of five trials. After retaining data from previous attempts, Qwen3.5-9B's success rate rose from 45% in the early stage to 84% in the later stage. The research did not demonstrate a fully autonomous self-replicating AI worm, but Anthropic predicts that models with 300 billion parameters or fewer may reach the required capability by the end of 2026. Currently, completing inference deployment on the tested hardware requires 28 to 51 hours (Qwen3.5-9B) or 71 to 150 hours (Gemma4-26B-A4B). It was also reported that Anthropic published a blog this week stating that small models may acquire self-replication capability "within months" and called for strengthening the security of networked devices.
Anthropic warns that small AI models may be capable of self-replicating on commercial IoT devices within a few months, thereby forming a distributed AI workload network that does not depend on data centers. Related experiments show that Qwen3.5-9B successfully installed and configured the software required to run Llama3.1-8B on eight of nine hardware types; GLM-5.2 reached all eight nodes of a controlled simulated network in four of five trials. After retaining data from previous attempts, Qwen3