什么是 GPT-4o？

Thinking Machines released TML-Interaction-Small 276B-A12B, a 276-billion-parameter MoE model with only 12 billion active parameters, achieving sub-200ms end-to-end latency, surpassing GPT-4o and Gemini 3.1-Flash in real-time voice interaction, time-aligned microturns, and visual proactivity, effectively eliminating standard VAD.

MiniMax-M3 has launched on OpenRouter — a frontier-class open-weight model supporting 1M-token context, agentic performance, and native multimodality (image & video), marking a major leap in long-context, autonomous-agent, and multi-modal AI capabilities.

谷歌发布 Gemma 4 QAT 模型，内存占用减少 3 倍，可在手机和笔记本电脑上运行。

FrontierCode 是一项新的代码质量评估基准，专注于衡量代码是否可合并，而非仅通过单元测试。

Martin Fowler analyzes flaws in metrics for evaluating AI tools, notes automation hasn't eliminated professions, open-source models are catching up faster to closed models, and hallucinated citations in AI-generated content.