Google released Gemini 2.0 Pro Experimental (2 million token context, tool calling, code execution), made Gemini 2.0 Flash generally available, and introduced Gemini 2.0 Flash-Lite as a cost-competitive tier matching the price and speed of Gemini 1.5 Flash while claiming superior performance. The release explicitly positions Google's models against DeepSeek's R1 reasoning models and their "relative steal" API pricing. Google is distributing these models across Vertex AI, Google AI Studio, and its consumer Gemini app, including a reasoning variant (Flash Thinking) to the app user base.
160 other tracked developments touching the same claims.