Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language can often provide a much richer learning medium for LLMs, compared with policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA (Genetic-Pareto), a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing one or more LLM prompts, GEPA samples system-level trajectories (e.g., reasoning, tool calls, and tool outputs) and reflects on them in natural language to diagnose problems, propose and test prompt updates, and combine complementary lessons from the Pareto frontier of its own attempts. As a result of GEPA's design, it can often turn even just a few rollouts into a large quality gain. Across four tasks, GEPA outperforms GRPO by 10% on average and by up to 20%, while using up to 35x fewer rollouts. GEPA also outperforms the leading prompt optimizer, MIPROv2, by over 10% across two LLMs, and demonstrates promising results as an inference-time search strategy for code optimization.

Melior Motion's planetary gearboxes are used in industrial robots with medium to high payload capacities. Schaeffler Group, an automotive and industrial supplier, signed an agreement to acquire all shares of Melior Motion for an undisclosed amount. With the acquisition of Melior Motion, Schaeffler adds more powerful gearboxes for higher payloads to its portfolio. The acquisition strengthens Schaeffler's robotics portfolio. The company began expanding its robotics presence in 2017 when it brought a strain-wave gearbox for use in a collaborative robot to production.

Toshiba Corporation has developed the world's most accurate highly versatile Visual Question Answering (VQA) AI, able to recognize not only people and objects, but also colors, shapes, appearances and background details in images. The AI overcomes the long-standing difficulty of answering questions on the positioning and appearance of people and objects, and has the ability to learn information required to handle a wide range of questions and answers. It can be applied to a wide range of purposes without any need for customization. In experiments using a public dataset comprising a large volume of images and data text, the VQA AI correctly answered 66.25% of questions without any pre-learning and 74.57% with pre-learning. For example, the AI can find a worker standing in a designated place by asking questions like, "is the person on a black mat?" which requires recognition of the individual, position, shape and color.

It's hard to believe how quickly the future is approaching. Waymo, formerly the Google Self-Driving Car Project, plans to launch its first commercial service for self-driving cars later this year, essentially a fully automated ride-hailing service covering about 100 square miles in Phoenix, Arizona. The company is called Navya, and its two self-driving cars currently on sale include the Autonom Shuttle and the smaller Autonom Cab. Both run on fully electric power, and while the top speed is 55 mph the vehicles are intended for urban environments only where maximum speeds reached are only around 30 mph. Navya says its vehicles have been designed for "last mile" travel, i.e. the final leg of a trip.