Google launches Gemini 3.7 Flash and cuts prices... but the Pro model is missing

At Google I/O in May, Google promised Gemini 3.5 Pro within a month, but three months later the model has not appeared, reportedly failing internal criteria on coding and reliability. Several key figures have left, including Jeff Dean (to cofound Discovery Loop) and Demis Hassabis (stepping back from operational leadership at DeepMind). On August 13, Google announced Gemini 3.7 Flash, claiming it outperforms its predecessor on benchmarks like FrontierCode (43.6% vs 34.4%), DeepSWE (65.3% vs 49.0%), WebDev Arena Elo (1588 vs 1538), GDP.pdf (34.0% vs 22.0%), and AutomationBench (30.4% vs 17.0%). It retains a 1-million-token context window and 64k output tokens. Until December 31, 2026, pricing is $0.75 per million input tokens and $3.75 per million output tokens, half the launch price of Gemini 3.6 Flash; the reduced rate applies retroactively to 3.6 Flash. From January 1, 2027, prices will rise to $1.50 and $7.50. Google doesn't claim top performance in math or reasoning but emphasizes the best intelligence-speed-cost tradeoff for daily work. The current high-end Gemini 3.1 Pro is deprecated, with Google admitting 3.7 Flash is better in almost everything. Gemini 3.5 Pro appears scrapped, as Google focuses on Gemini 4, which is in pre-training with no release date, possibly this autumn. Google's strategy may be sound since most professional uses don't need frontier models, and Google controls the full stack, enabling aggressive pricing, but its number one position is now contested by Anthropic and OpenAI.