Don't discontinue Gemini 2.5 Flash (discuss.ai.google.dev)

A recent appeal from users highlights the critical need for Google to retain the Gemini 2.5 Flash model, which is deemed superior for specific workflows compared to its successor, Gemini 3 Flash. Users report that despite attempts to optimize performance through new prompting guidelines, 2.5 Flash consistently outperforms 3 Flash and its variant, 3.1 Flash Lite, particularly in low-latency applications essential for voice agents. The significant difference in response times—300-400ms for 2.5 Flash versus 600-800ms for the newer models—demonstrates its practicality and reliability in real-world scenarios. This plea underscores a larger concern within the AI/ML community about model discontinuation and its impact on existing applications. The reliance on 2.5 Flash for efficient performance, especially in regions like Australia where alternatives are limited, indicates that discontinuing the model could disrupt numerous services. As organizations strive to enhance user experiences through AI, maintaining access to high-performing models like 2.5 Flash becomes essential for ensuring continued innovation and applicability in AI-driven solutions.