Agent
Google's Gemini models and the Gemini CLI agent, used for firmware, bring-up and bench work.
Ranked by points weighed against age. New entries with early votes rise.
15 d ago
Artificial Analysis launched AA-Robotics v0.1, a benchmark in which language models in coding-agent harnesses control a Hiwonder NexArm robot arm. Models in Codex, Claude Code and Antigravity write code for 8 zero-shot tabletop tasks, watching through 4 cameras. GPT-6 Astra led with 69.4 and Claude Opus 5.5 scored 48.8. Models took about 30 minutes a task; people averaged 20 seconds.
3 mo ago
Anthropic's red team had Claude, GPT, Gemini and other models drive robots, mostly in simulation plus a real Unitree Go2. The team reports that Opus 4.6 misjudged where a trash can was and the Go2 dragged it. Grok 4.1 Fast headed for a table reflected in a glass door, and no model completed an office loop on vision alone.
4 mo ago
Zhihan Zhang and colleagues, mostly at the University of Washington, published Embedded Arena, a benchmark that scores agents' model and firmware edits on microcontrollers. Boards include a MAX78000, an ESP32-S3 and an STM32N6. The paper reports Claude Opus 4.7 and Gemini 3.1 Pro failing every deployment without hardware feedback; with it, the first success came within 3 iterations.
5 mo ago
Parsa Ghasemi published the slides and demo prompts from a guest lecture to mechanical engineering students at Cal Poly Pomona. It covers installing Gemini CLI and Claude Code, writing prompts for runnable embedded code and catching wrong pins, invented library calls and unit errors. The live demo codes a PID position controller for a Teensy 4.1. We found no recording.
5 mo ago
Dimdim had Gemini 3.1 Pro write the Verilog for an audio reclocker on a MAX 10 10M08 board. He pasted code from the chat into Quartus. The design passes I2S through a dual-clock FIFO, adds DSD and S/PDIF input and buffers in PSRAM. Gemini, Claude Sonnet 4.6 and Claude Opus 4.7 all failed to make a PRBS-15 test work.
6 mo ago
lego-robot-mcp has Claude Code drive a LEGO robot with two wheels, a gripper and an arm on a Raspberry Pi 4 with a Build HAT. Claude sees through a Pi Camera on the robot and an overhead iPhone, and a Gemini Robotics-ER model checks frames after each move. The README has no photos or video of the robot.
6 mo ago
Seif Salama built an Ackermann-steered robot that runs one ROS 2 Humble stack in Gazebo Harmonic and on a Raspberry Pi car. A custom ros2_control interface drives the motors over UART and the steering servo over I2C. A Gemini or Ollama agent takes spoken commands. We could not tell whether the README video shows the physical robot.
