A growing number of developers are pleading with Google to keep Gemini 2.5 Flash available, citing significant performance advantages over newer models like Gemini 3 Flash and 3.1 Flash Lite. Users report that despite attempts to optimize prompts for the newer versions, Gemini 2.5 Flash consistently outperforms them in internal benchmarks, particularly in latency-sensitive applications.

One primary use case highlighted is voice agents, where Gemini 2.5 Flash achieves completion times of 300-400ms. In contrast, the newer Gemini 3.5 Flash model reportedly delivers 600-700ms completions, making it unsuitable for real-time voice interactions. This latency difference is further exacerbated by deployment issues; Gemini 2.5 Flash is deployed in Australia, offering near real-time performance, whereas 3.5 Flash has no Australian deployment, pushing latency closer to 700-800ms.

Developers express concern that discontinuing Gemini 2.5 Flash would represent a substantial loss for specific workflows, especially those requiring low-latency responses and high-quality output. The model is considered the best for all-around usage and is critical for deployments in regions like Australia, where no other comparable low-latency model is available.