🧪 Models of the week: a data-driven comparison
This week, the main question isn't "who's smarter" but "who can you trust with your desktop". OpenAI's GPT-6.1 Terra is clearly aiming at desktop agents: it's being tested on OSWorld 2.1, where the model has to build a 3D model from a technical drawing and complete multi-step tasks in everyday applications. Its strength here is precisely working with interfaces, not chat. The weak spot: the more complex the scenario, the costlier a mistake — and there's no public data on speed and pricing in the news.
Claude #models #benchmarks #AI #LLM #neuralnetworks
🚀 Subscribe to @ai_release1 — AI releases, tests, and comparisons every week.
Source: t.me · post in Telegram