Another huge week of AI and robotics news I summarized everything...

I summarized everything from Google, Nvidia, OpenAI, Microsoft, Agility Robotics, LimX Dynamics, Boston Dynamics, xAI, Anthropic, and more.
Here's everything you need to know and how to make sense out of it:
Ranked #1 on LMArena, it supports consistent multi-turn image editing with text prompts, and can even blend photos
Now available for free and paid Gemini users
It uses a 2,560-core Blackwell GPU, 96 fifth-generation Tensor cores, and 128GB of memory to deliver up to 2,070 FP4 teraflops of AI compute
—gpt-realtime speech-to-speech model with remote MCPs and image support
—Realtime API in general availability
—New features for Codex, including an extension to run in IDEs, code reviews, and more
The former can generate a minute of speech in a second, while the latter specializes in everyday text queries
Both are being tested yet, with MAI-1's test version ranking #13 on LMArena

Currently in training, the neural network will teach whole-body skills to Agility's humanoids while enabling them to handle disturbances and heavy objects
In the clip, the robot can be seen recovering from different surfaces as well as retaining its position/posture in response to external impacts
The flips build on RL pipelines from the Robotics and AI Institute, with Spot training through countless trial-and-error cycles until it executes the move
The model uses a new lightweight architecture for cost efficiency and is available for free on major IDEs
xAI also open-sourced Grok 2.5 this week
Currently, the company is testing it as a research preview with 1,000 Max plan users to gather real-world insights and iron out potential safety issues
It beats GPT-5, Deepseek-V3, and Google on key benchmarks across 23 major business languages
Also includes ‘Deep Translation’ to double-check high-stakes content

It combines a motion capture setup with an AI-based planner and whole-body controller to execute strikes with sub-sec latency
It generates film-grade, audio-driven human animation with advanced motion and environment control via instruction
Fully open-source
The company currently sells a dual-arm robot called R1 and has plans to launch a legged humanoid later this year
It was founded by a Stanford postdoc and former Waymo engineer
> Software Test Eng
> End of Line Lead
> Equipment Eng
> Technician
> Manufacturing Eng
Apply here: figure.ai/careers

We're hiring for 35 roles in BotQ - our robot manufacturing division at Figure:
→ Software Test Eng
→ Electrical Test Eng
→ End of Line Lead
→ Warehouse Lead
→ Equipment Eng
→ Technician
→ MFG Eng
I share the latest research every week, so follow me @adcock_brett for more.
If you found this valuable, consider a like/retweet to spread the word.

I summarized everything from Google, Nvidia, OpenAI, Microsoft, Agility Robotics, LimX Dynamics, Boston Dynamics, xAI, Anthropic, and more.
Here's everything you need to know and how to make sense out of it:



