SemiAnalysis 发布 GDPx 基准：最强模型仅得 12%
SemiAnalysis 搞了个新基准 GDPx，最强模型只拿 12%，里面 Gemini Carbon 在智能体任务上表现挺亮眼，可以看看。
SemiAnalysis 发布 GDPx 基准，用于衡量模型在高价值领域的智能体能力。该基准达到前沿难度，目前最好的模型只拿到 12%（7 题中答对 1 题）。Elvis Saravia 转评提到，Gemini 新版本 Carbon 在协调与持久性任务上表现明显提升。这类脱离标准基准的测试正在揭示各模型能力上的差异。
Recommended read. Always great to have new ways to measure the impact of frontier intelligence. What strikes me the most is how much better Gemini (now with Carbon) has gotten. Don't count out Google just yet. As we move away from standard benchmarks, you will start to see all sorts of interesting results and surprises that tell a different story about these models' capabilities and how they could unlock economic value. Carbon, in particular, is a very strong model for coordination and persistence, which is becoming increasingly clear in some of these new benchmarks. SemiAnalysis @SemiAnalysis_ Today we’re releasing GDPx, a benchmark for measuring agentic capabilities across the most valuable domains. GDPx is a true frontier-level benchmark: even the best model scores just 12% (1/7)🧵 🔗 View Quoted Tweet 💬 20 🔄 2 ❤️ 43 👀 7160 📊 18 ⚡
