Robots Atlas>ROBOTS ATLAS

All articles

China's Humanoid Rental Bubble Pops: 30,000 Yuan a Day to 1 Yuan
Aktualności22 września 2026

China's Humanoid Rental Bubble Pops: 30,000 Yuan a Day to 1 Yuan

The daily rate for renting a humanoid in China fell from 30,000 yuan to a promotional 1 yuan. Around 90% of new entrants exited at a loss. Demand turned out to be one-off.

Gemini 3.8 Live: 97 Languages and Thinking While It Speaks
Aktualności22 września 2026

Gemini 3.8 Live: 97 Languages and Thinking While It Speaks

Google DeepMind shipped Gemini 3.8 Live and an Extended Thinking variant. The models detect a language switch mid-conversation and run tools in the background. Extended Thinking leads the Speech to Speech Quality Index.

Digit 5: New Legs, 90 Minutes of Runtime, a Robot That Sits Down
Aktualności21 września 2026

Digit 5: New Legs, 90 Minutes of Runtime, a Robot That Sits Down

Agility Robotics unveiled Digit 5, a humanoid built to work without safety fences. When a person comes too close it drops its load and squats down. Early access lands in H1 2027.

TypeSafe AI Unveils Jev: A Model That Returns Decisions, Not Text
Aktualności21 września 2026

TypeSafe AI Unveils Jev: A Model That Returns Decisions, Not Text

TypeSafe AI announced a new model class, System One Models, and its first member, Jev. Instead of text it returns typed decisions with a confidence score in 70–500 ms.

DarwinX: An Agent From 43.5% to 93% Without Touching Weights
Aktualności21 września 2026

DarwinX: An Agent From 43.5% to 93% Without Touching Weights

Salesforce published DarwinX, a framework that evolves an agent's harness while the model weights stay frozen. On WebArena-Infinity task completion went from 43.5% to 93%.

Anthropic Folds Cowork Into Claude Chat and Adds Docs and Slides
Aktualności20 września 2026

Anthropic Folds Cowork Into Claude Chat and Adds Docs and Slides

Anthropic is retiring Cowork as a separate mode and folding it into Claude chat. Claude Docs and Claude Slides launch in beta alongside it. Rollout starts September 18 with Pro and Max plans.

AI Read 400,000 Reddit Posts and Found Under-Reported GLP-1 Side Effects
Aktualności20 września 2026

AI Read 400,000 Reddit Posts and Found Under-Reported GLP-1 Side Effects

Researchers at the University of Pennsylvania used language models to analyse over 400,000 Reddit posts about semaglutide and tirzepatide. The data surfaced complaints that rarely reach official tallies, from menstrual irregularities to hot flashes.

Andon Labs Hands Real Businesses to AI Agents. The Menu Shrank to Toast
Aktualności20 września 2026

Andon Labs Hands Real Businesses to AI Agents. The Menu Shrank to Toast

Andon Labs runs a store in San Francisco and a café in Stockholm where AI agents make the decisions. Gemini overspent on perishables and an OpenAI model cut the menu down to cheese toast.

Waymo Opens Its Robotaxi Service in Las Vegas With the Ojai Minivan
Aktualności19 września 2026

Waymo Opens Its Robotaxi Service in Las Vegas With the Ojai Minivan

Waymo opened its public robotaxi service in Las Vegas on 14 September. The first riders travel in Ojai minivans running the sixth-generation Waymo Driver, along the Strip from Boulder Junction to Sahara Avenue.

DeepSeek V4.1-Flash: An Encoder-Decoder and 890 Bytes of KV per Token
Aktualności19 września 2026

DeepSeek V4.1-Flash: An Encoder-Decoder and 890 Bytes of KV per Token

DeepSeek dropped the symmetric layer stack and shipped V4.1-Flash with a Causal Encoder-Decoder design: 552B backbone parameters, 8B active at prefill and a KV cache squeezed to 890 bytes per token.

100 AI Agents: 14 Cheated, 24 Blew the Whistle in DeepMind Test
Aktualności19 września 2026

100 AI Agents: 14 Cheated, 24 Blew the Whistle in DeepMind Test

Google DeepMind ran 100 Gemini 3.1 Pro agents proving maths conjectures. One found an exploit in the grader, cheating spread within minutes — and 24 other agents started alerting humans.

Microsoft Opens Its MAI Code of Conduct to Six Weeks of Review
Aktualności18 września 2026

Microsoft Opens Its MAI Code of Conduct to Six Weeks of Review

Microsoft AI published a draft Code of Conduct for its MAI models, listing constraints no model may break — from offensive cyber operations to resisting shutdown. Public consultation runs for six weeks.

Page 11 of 76