Google's Cheaper AI Model Leaps Ahead of Flagship

Google has released Gemini 3.7 Flash, its latest AI model that outperforms its predecessor in coding benchmarks by large margins. The new model scores 43.6% against 34.4% on FrontierCode 1.1 and 65.3% against 49.0% on DeepSWE v1.1.

The release comes as Google continues to push out cheaper models at a faster rate, while its flagship Gemini 3.5 Pro remains months behind schedule. The company has priced the new model at $0.75 per million input tokens and $3.75 for output until the end of December, after which both prices will double.

Google also claims to have added safeguards against cyber offense and misuse in line with its Frontier Safety commitments. However, the release of Gemini 3.7 Flash has not been without controversy, with Axios reporting that Google could skip the 3.5 Pro model entirely and go straight to Gemini 4 Pro.

The delay in releasing Gemini 3.5 Pro has reportedly had a negative impact on DeepMind morale, particularly given that rival labs are hiring at this time.