The AI model that just scored 65% on DeepSWE isn’t the one Google promised.

Google launched Gemini 3.7 Flash, a new AI model, but it's not the Gemini 3.5 Pro model they previously promised. This new model scored 65% on DeepSWE, a benchmark test. It's unclear what caused the delay or discrepancy. Engineers should be aware of the confusion and verify model details before implementation. The Gemini 3.7 Flash model is now available for testing.