Agentic Coding

Meituan released LongCat-2.5, a 1.6T-parameter native multimodal model with roughly 48B active parameters per token. It was trained with long-horizon reinforcement learning for agentic coding and optimized for W4A4 inference on domestic compute.

Model Release • Pandaily

Productivity

According to MIT Technology Review, an AI coding agent plans, generates code, runs tests and repeats to correct errors, and a single task can use 10 to 50 times more tokens than a basic chat prompt.

Google released Gemini 4 Carbon, a frontier model it says delivers significant gains in coding, agentic tasks, and cybersecurity. Google claims Carbon tops the Artificial Analysis Intelligence Index v5.0, ahead of Claude Opus 5.5 and GPT-6 Astra.

AI Models • The Verge AI