415.tech
AI & tech, from the frontlines of Silicon Valley
Gemini 3.7 Flash undercuts its three-week-old predecessor by 50% on price

Gemini 3.7 Flash undercuts its three-week-old predecessor by 50% on price

Google shipped Gemini 3.7 Flash at $0.75 per million input tokens and $3.75 per million output tokens — half what Gemini 3.6 Flash cost at launch three weeks ago — with FrontierCode rising from 34.4% to 43.6% and DeepSWE from 49.0% to 65.3%. Google's own numbers put it ahead of Claude Sonnet 5 and GPT-5.6 Terra, so coding agents now have a cheaper frontier-adjacent option through the API, AI Studio, and Antigravity. Launch pricing holds through year-end, and the three-week model cadence means the tier itself keeps moving.

Source: the-decoder.com ↗

Post on XEmail

Google's benchmarks also show gains in web development, document comprehension, and business process automation.

Google

Why this matters

  • → Coding benchmarks jump 26% in three weeks—frontier-adjacent performance now at half the prior price
  • → 50% price cut on launch day signals aggressive model-tier compression and faster release cadence
  • → Cheaper option for building coding agents compresses Claude and OpenAI's margin at the frontier
Speed meets price