Gemini 3.7 Flash
Points and comments are a snapshot, not live.
Google launches Gemini 3.7 Flash, claiming major coding gains at half price.
Google released Gemini 3.7 Flash, its most intelligent workhorse model for coding and agents, just three weeks after 3.6 Flash. It shows strong improvements on benchmarks: FrontierCode 1.1 Main (43.6% vs 34.4%), DeepSWE v1.1 (65.3% vs 49.0%), and WebDev Arena Elo (1588 vs 1538). Introductory pricing through 2026 is $0.75/1M input tokens and $3.75/1M output tokens. The model powers Gemini Spark for Google AI Pro and Ultra subscribers in over 160 countries.
What commenters are saying
Commenters are skeptical, noting Google has fallen behind OpenAI (GPT 5.6 Terra) and Anthropic (Claude Sonnet 5) on frontier benchmarks. Many compare unfavorably to DeepSeek V4 Flash/Pro, which they say is 13-26x cheaper with comparable intelligence. A few defend the model on latency, uptime, and scale: "the only thing that works at scale is gemini flash." Others point out DeepSeek already announced a price hike.