Google shipped Gemini 3 Flash on Wednesday, immediately making it the default model across the Gemini app and AI Mode in Search. The model slots in below Gemini 3 Pro in the lineup but matches or beats it on several benchmarks, which raises an obvious question: why pay more for Pro?
The company positions Flash as the "workhorse model," per Tulsee Doshi, senior director of Gemini product. It costs $0.50 per million input tokens and $3 per million output, compared to Pro's $2 and $12 respectively. Google claims it outperforms the previous Gemini 2.5 Pro while running three times faster, citing third-party benchmarks from Artificial Analysis.
On Google's own numbers, Flash scores 90.4% on GPQA Diamond and 33.7% on Humanity's Last Exam without tools. For context, Gemini 3 Pro hit 37.5% on HLE while GPT-5.2 scored 34.5%, though all these figures come from the companies themselves. Flash actually beat Pro on SWE-bench Verified for coding tasks (78%), which seems backwards until you consider the benchmarks might not capture what makes Pro worth the premium.
JetBrains, Figma, Cursor, and Harvey are already using Flash in production. Google says the API has been processing over a trillion tokens daily since launching Gemini 3 Pro last month.
The Bottom Line: Google now defaults users to a model that nearly matches Pro performance at a quarter of the price, betting volume matters more than margin.
QUICK FACTS
- API pricing: $0.50/1M input tokens, $3/1M output tokens
- Released: December 17, 2025
- GPQA Diamond: 90.4% (Google-reported)
- SWE-bench Verified: 78% (Google-reported)
- Context window: 1M input tokens, 65K output tokens
- Knowledge cutoff: January 2025
