According to 9to5Google, Google released its next-generation large model Gemini 4 Hydrogen this week. The company has dropped the "Flash" naming and positioned the model as "bringing frontier intelligence to more users." The model natively supports million-level token input and output. Compared with the previous generation released in September, the default tier completes tasks 2.5 times faster, reduces fabricated statements by more than a factor of five, and can adaptively handle tool call failures or empty retrieval results.
