OpenAI, Anthropic and Google’s 2025 AI arms race dissected

Shanghai-based outlet The Paper published a sprawling year‑end analysis of the "AI big three"—OpenAI, Anthropic and Google—framing 2025 as a bruising arms race of model launches, valuations and ecosystem plays. The piece credits OpenAI’s GPT‑5.2, with its huge context window and agentic capabilities, as a late‑year comeback that helped it reclaim a perceived technical edge after Google’s Gemini 3 briefly dominated benchmarks and mindshare. Anthropic’s Claude 4.5 is portrayed as the workhorse of long-running autonomous tasks and enterprise workflows, buoyed by a massive funding round that pushed its valuation into the top tier of private companies, while Google’s Gemini 3 is cast as the champion of native multimodality and aggressive product integration across Search, Android and Workspace. The article also zooms out to highlight how Microsoft, Meta, Apple, Mistral and Elon Musk’s xAI shifted strategies—Microsoft hedging beyond OpenAI by embracing Claude, Meta doubling down on open models, Apple quietly baking on-device AI into its platforms, and European champion Mistral riding a blockbuster raise. It reads less like simple hype and more like a sober Chinese-language scorecard on how power, capital and technical direction are consolidating in the global model race heading into 2026.